Inflammation‐Driven Lymphoid Structures: Organization, Function, and Clinical Impact Across Autoimmunity, Cancer, and Checkpoint Toxicity Authors: Marie Frutoso, Emie Delmas, Rudolf Corty, Eleonore Bettacchioli, Divi Cornec, Sophie Hillion, Soizic Garaud Published: 2026 ABSTRACT ABSTRACTLymphoid structures (LS) arising in nonlymphoid tissues are increasingly recognized as active immune niches rather than simple consequences of chronic inflammation. In this review, we propose a framework that distinguishes inflammatory lymphoid structures (ILS) from mature tertiary lymphoid structures (mTLS), while acknowledging that these entities may form a continuum in some settings and represent distinct inflammatory states in others. We discuss how LS develop across autoimmunity, cancer, transplantation, immune‐related adverse events, and aging, and how differences in architecture, stromal organization, antigen persistence, and tolerance checkpoints shape their biological outputs. Whereas mTLS are typically associated with follicular organization, germinal center‐like reactions, and local clonal diversification, ILS may instead sustain inflammation through extrafollicular activation, survival niche formation, and tissue‐adapted immune specialization. Importantly, the functional impact of LS is highly context dependent: they may promote pathogenic autoreactivity in autoimmune disease, support protective antitumor immunity in cancer, or acquire regulatory or suppressive features depending on the microenvironment. We also review current therapeutic strategies targeting LS and discuss how spatial transcriptomics and artificial intelligence may refine their detection, classification, and functional interpretation. Altogether, this perspective supports viewing LS as heterogeneous tissue immune niches rather than a binary TLS category, with major implications for pathogenesis, biomarker development, and therapeutic intervention. Introduction Lymphoid structures (LS) arising within nonlymphoid tissues have been described for decades under multiple designations, including ectopic germinal centers, ectopic lymphoid follicles, tertiary lymphoid structures (TLS), or ectopic lymphoid structures. These structures emerge across a broad spectrum of chronic inflammatory contexts, including cancer, autoimmune diseases, persistent infections, allo‐immunity, and aging. While initially considered epiphenomena of ongoing inflammation, they are now increasingly recognized as spatially organized immune niches capable of sustaining local adaptive responses and reshaping immune dynamics directly within tissues. Their functional impact, however, varies by context. In cancer, LS are frequently associated with improved prognosis and enhanced responses to immune checkpoint blockade, consistent with effective antitumor clonal expansion [1, 2, 3]. In autoimmune diseases, LS often correlate with autoantibody diversification, epitope spreading, and tissue damage, suggesting local amplification of autoreactive clones [4, 5]. Immune‐related adverse events induced by checkpoint inhibitors illustrate another scenario in which therapy‐driven immune activation promotes de novo lymphoid organization within nonlymphoid tissues, potentially unleashing previously constrained clones and altering tolerance thresholds. A central issue in the field is how these lesions should be classified. Several studies support a morphology‐based view of lymphoid neogenesis as a progressive increase in organization, from loose lymphoid aggregates to follicle‐like immature TLS and, ultimately, germinal center‐containing mature TLS (mTLS) [6, 7, 8]. In this framework, the earliest lesions consist of poorly organized lymphocytic aggregates lacking follicular dendritic cell networks, followed by more structured follicle‐like or immature TLS containing segregated T‐ and B‐cell areas and FDCs, and finally mTLS harboring germinal center (GC) reactions. This framework captures one important mode of LS development. However, the precise dynamics of this maturation process remain incompletely resolved, particularly in human autoimmunity. Less organized lesions could therefore not be viewed exclusively as transient or incomplete precursors of mTLS. Rather, these structures may follow different trajectories: some may progressively acquire higher levels of stromal specialization, compartmentalization, and repertoire selection, whereas others may persist as distinctly organized and functionally active inflammatory niches. To accommodate both possibilities, we use in this review the term inflammatory lymphoid structures (ILS) to designate LS whose architecture differs from that of highly organized mTLS. Although ILS may overlap morphologically with lesions referred to as early or immature TLS, the term does not presuppose a mandatory evolution toward mature TLS and leaves open the possibility that these structures may also exist as independent inflammatory immune states. A related question concerns the role of antigen persistence and clonal selection in shaping these structures. In secondary lymphoid organs (SLOs), antigen availability, cell activation and tolerance checkpoints are tightly coordinated to regulate the immune repertoire, ensuring efficient pathogen clearance while limiting autoreactivity. By contrast, LS arising in inflamed tissues develop in environments characterized by sustained inflammation, stromal remodeling, altered vascularization, and inflammatory cytokine gradients. In such contexts, the balance between clonal expansion, diversification, and tolerance control may be reconfigured (Table 1). TABLE 1ILS's specific features in comparison of SLO and TLS.StructureCellular organizationTriggerFunctional outputDominant antigenic contextRepertoire dynamicsTolerance controlReferencesSLO Fully structured compartmentalized T/B zones, FDC network, GC formation: Segregated DZ/LZ Clonal proliferation Developmental (LTβR axis, embryonic stromal cues) Homeostatic immunity High affinity antibodies Memory B cells generation Acute infection Vaccination Antigen‐driven clonal expansion under strict; regulated SHM and selection Strong peripheral checkpoints Stringent negative selection Regulated Tfh help [9, 10]LS ILS Immature TLS ELS Partial organization; variable stromal environment No clear FDC structure Polyclonal diversification Acute/chronic inflammation Auto‐Ag Inflammatory amplification Local immune modulation Autoantibody production? Chronic self‐Ags Chronic infection No Ag Early clonal expansion; repertoire skewing No/Low selection EF differentiation type like Low level? Inflammatory cytokines (BAFF, IFN‐I, IL‐6) Lower activation thresholds [7, 11, 12]Mature TLS Compartmentalized FDC networks; vascular remodeling Clonal proliferation? Sustained chronic inflammation with persistent antigenic stimulationLocal adaptive immunity autoantibody production or tumor immunity regulation Tumor Ag Neo Ag Specific infection Local affinity maturation; clonal diversification; potential epitope spreading Context‐dependent tolerance modulation Ag persistence Local modulation Tfh regulation [1, 2, 3] ILS's specific features in comparison of SLO and TLS. Fully structured compartmentalized T/B zones, FDC network, GC formation: Segregated DZ/LZ Clonal proliferation Homeostatic immunity High affinity antibodies Memory B cells generation Acute infection Vaccination Strong peripheral checkpoints Stringent negative selection Regulated Tfh help ILS Immature TLS ELS Partial organization; variable stromal environment No clear FDC structure Polyclonal diversification Acute/chronic inflammation Auto‐Ag Inflammatory amplification Local immune modulation Autoantibody production? Chronic self‐Ags Chronic infection No Ag Early clonal expansion; repertoire skewing No/Low selection EF differentiation type like Low level? Inflammatory cytokines (BAFF, IFN‐I, IL‐6) Lower activation thresholds Compartmentalized FDC networks; vascular remodeling Clonal proliferation? Tumor Ag Neo Ag Specific infection Context‐dependent tolerance modulation Ag persistence Local modulation Tfh regulation Although persistent self‐antigens, tumor neoantigens, or chronic pathogen‐derived antigens can promote lymphoid organization and shape local evolving B‐ and T‐cell repertoire, LS have also been observed in situations where a dominant antigenic driver is not clearly identified. Aging provides a compelling example. Progressive stromal remodeling, chronic low‐grade inflammation, and altered chemokine expression may favor immune cell organization and ectopic organization even in the absence of a defined antigen [13, 14, 15]. In such settings, lymphoid assembly may be sustained by systemic inflammatory, age‐related changes in stromal cell plasticity, or homeostatic dysregulation. In this review, we first discuss the structural and cellular characteristics of LS arising in inflamed nonlymphoid tissues across distinct inflammatory contexts, including rheumatoid arthritis, Sjögren's disease, systemic lupus erythematosus, IgA nephropathy, ANCA‐associated vasculitis, transplantation, immune‐related adverse events, and aging. We then examine how these different forms of lymphoid organization (ILS and organized mTLS) may contribute to disease pathogenesis through local amplification of immune responses, survival niche formation, cellular specialization, and context‐dependent regulatory or protective functions. Next, we review current therapeutic strategies aiming at modulating LS formation or function, either to restrain pathogenic inflammation or to enhance beneficial local immunity. Finally, we discuss how emerging technologies, particularly spatial transcriptomics, high multiplexed imaging and artificial intelligence approaches, may refine the detection, classification, and functional interpretation of ILS and mTLS in tissues. Through this presentation, we aim to highlight both shared principles and unresolved questions regarding the biology and clinical relevance of inflammation‐driven lymphoid structures. Structural and Cellular Characteristics of LS Across Inflammatory Contexts Rheumatoid Arthritis In rheumatoid arthritis (RA), the synovium is the main target tissue of chronic autoimmune inflammation and displays a broad spectrum of LS [16]. Histological heterogeneity was already evident in early synovial studies [17], and was later formalized into the three recurrent patterns: diffuse infiltrates, T/B aggregates lacking follicular dendritic cells (ILS), and mature follicular structures with GC (mTLS) [18]. In a series of 64 RA synovial biopsies, mTLS were identified in 23.4% of patients [19], and broader estimates indicate that LS structures are present in at least 40% of RA synovia, although only 10%–25% show features of functional GC. In an early‐RA cohort of 144 patients, Humby et al. [20] identified three major pathotypes: lympho‐myeloid, diffuse‐myeloid, and pauci‐immune/fibroid pathotypes that were associated with both 6‐month response to disease‐modifying antirheumatic drugs and subsequent radiographic progression. The lympho‐myeloid pathotype showed the highest levels of disease activity and the highest frequency of rheumatoid factor and anti‐citrullinated protein autoantibodies seropositivity. Notably, it was also the only pathotype associated with LS (mostly ILS), consistent with its enrichment in genes involved in lymphoid organization and immune cell recruitment, including CXCL13, CCL19, LTβ together with inflammatory mediators such as IL‐6, CSF2, and MMP10. The B‐cell compartment is particularly prominent along this ILS/mTLS spectrum. In mTLS with GC‐like organization, B cells are accompanied by plasmablasts (PB) and plasma cells (PC), and these structures are functionally supporting class‐switch recombination and ongoing production of autoantibodies [21]. Synovial PB and PC analyses demonstrated selected local clonal expansion and accumulation of PC clones within the joint [22], supporting in situ differentiation. Recombinant monoclonal antibodies derived from B cells differentiating inside RA synovial LS recognized citrullinated and NET‐related antigens, providing direct evidence that these structures can sustain local autoreactive responses [23]. Recent single‐cell studies identified a synovium‐enriched NR4A1+ activated B‐cell population with a transcriptional overlap with germinal‐center light‐zone B cells, progressive somatic hypermutation, and clonal sharing with local plasma cells, consistent with ongoing antigen‐driven activation inside synovial LS [24]. These NR4A1+ B cells co‐express LTα, LTβ, and IL‐6, suggesting that they may help sustain synovial LS. More broadly, single‐cell RNA‐seq and mass cytometry identified naïve B cells, memory B cells, CD11c+ (ITGAX) age‐associated B cells (ABC), and PB with frequencies varying according to the histological nature of infiltrate. ABCs were found almost exclusively in leukocyte‐rich RA biopsies and were associated with the expansion of CD4+PD1+ICOS+ T peripheral helper/T follicular helper‐like populations linked to CXCL13‐rich helper programs [25]. Immunofluorescence further suggested that CD11c+ B cells are more frequently observed in ILS than mTLS. A 2024 paired synovium/blood transcriptional and repertoire study confirmed eight different B cell clusters and two PB clusters, including the previously identified NR4A1+ activated B cells, naïve B cell clusters, memory, and FcRL5+ TBX21+ABCs [26]. By matching BCR sequences to transcriptionally defined B cell states, this study demonstrated that clonally expanded PC in synovium shared BCRs with activated B cells, ABCs, and memory B cells, indicating that multiple synovial B‐cell states can feed the PC pool and sustain locally produced antibodies. The same study also predicted specific T peripheral helper/B‐cell interactions through IFN/IFNGR, LTA/TNFRSF14/HVEM, and BTLA/HVEM pathways, and showed that T peripheral helper cells uniquely coproduced IFNγ and CXCL13, providing an interesting model for how ILS may be maintained or pushed toward more organized mTLS. Beyond local immune activation, B cells may also contribute directly to structural damage, as memory B cells can produce RANKL and thereby link B‐cell activation to bone erosion [27], while B cells can inhibit osteoblast differentiation and bone formation [28]. Finally, spatial transcriptomic analysis of treatment‐refractory RA patients showed that baseline synovia can contain distinct “immuneT” and “immuneP” niches enriched in T/B‐cell infiltrates and PC aggregates. After triple‐therapy disease‐modifying antirheumatic drugs or adalimumab treatment, both remitting and non‐remitting patients showed depletion of the immune compartment. However, non‐remitting patients developed expansion of fibrogenic vascular niches driven by TGFβ signaling. These data suggest that reducing the immune compartment alone may not be sufficient to achieve clinical and open new avenues for understanding the link between preexisting ILS/TLS programs and fibroblast‐dominated refractory remodeling [29]. Sjögren's Disease In Sjögren's Disease (SjD), diagnosis is based on the 2016 ACR/EULAR classification criteria, which assign weighted scores to objective oral dryness, ocular dryness, anti‐SSA autoAbs, and focal lymphocytic sialadenitis in minor salivary glands [30]. The latter is assessed by the focus score (FS), defined as an aggregate of more than 50 mononuclear cells adjacent to ducts in a 4 mm2 tissue [31]. FS ranges from mild lesions (1 < FS < 2) to intermediate (2 < FS < 3) and severe lesions (FS > 3), with marked variation in infiltrate composition. In particular, severe lesions are characterized by increased infiltration of B cells and macrophages, along with a reduced proportion of helper T cells [32]. Severe lesions are also more frequently associated with mTLS and with an increased risk of lymphoma development [32, 33, 34]. Nonetheless, these represent less than 20% of LS [35], and fewer than 5% of SjD patients develop lymphoma [36] suggesting that ILS is the common form of lymphoid organization in SjD. Importantly, B cells in minor salivary glands are predominantly memory or activated cells [12, 37, 38]. Although some memory subsets within these LS express transcripts of LTβ and IL‐6 [37], two molecules involved in spontaneous GC formation in SLE [39], the relative paucity of naïve B‐cell subsets raises questions about the local priming of naïve B cells, which normally occurs in SLOs. A recent study further characterized salivary gland LS with or without mTLS by directly comparing their composition with SLOs. Interestingly, genes expressed in tissues with mTLS overlapped with those of tonsillar GC (Lymphotoxin β, CXCL13, TNFRSF13C, IL‐7, CXCR5, and CCR7), although at lower expression levels. Some of these genes were also detected in ILS (CCR7, ICOS). Remarkably, both LS (mTLS and ILS) displayed an inflammatory signature that was completely absent from tonsillar GC (IFNγ, TNFα, FASLG, CXCL10, and CXCL12) [40]. Finally, although most SjD patients display minor salivary gland infiltration, a subset of patients lacks detectable anti‐SSA antibodies. Increasing evidence suggests that SSA‐positive and SSA‐negative patients represent biologically distinct disease endotypes [41, 42], raising the possibility that the balance between ILS and mTLS may also differ across SjD subgroups. Overall, these observations support the view that salivary gland LS in SjD do not represent a uniform entity, but instead encompass at least two organizational states: predominant ILS, which may function as inflammatory lymphoid niches with a distinct mode of organization, and a smaller subset of more mTLS with GC‐like features more consistent with local B‐cell activation and selection (Figure 1). Organization of lymphoid structures in patient with Sjögren's Disease: One section of FFPE block of minor salivary glands from one SjD patient was cut and processed by Imaging Mass Cytometry. Images highlight the presence of LS in the minor salivary gland. Both ILS and mTLS could be observed, and the different markers enable to classify the organization of (A) an ILS structure and (B) a mTLS structure. Lupus Nephritis Lupus nephritis (LN) is often characterized by rapidly progressive glomerulonephritis, but the clearest evidence for pathogenic lymphoid organization within the kidney tissue comes from the tubule‐interstitium rather than the glomerulus [43]. Human biopsy studies showed early that intrarenal B‐cell infiltrates are heterogeneous and span a continuum from scattered B cells to organized LS with increasing structural complexity. In 46%–57% of LN biopsies, tubulointerstitial infiltrates are organized into T/B cell‐segregated ILS whereas fewer than 10% of biopsies were reported to contain structures consistent with GC‐like mTLS [44, 45]. However, the frequency of GC‐like mTLS varies across studies, likely reflecting the lack of a consensual definition; by example, no GC‐like structures were identified in a series of 192 LN cases [46]. Despite this limited degree of structural maturation, these renal infiltrates are not immunologically inert and were associated with intrarenal B‐cell clonal expansion [47] and ongoing somatic hypermutation, indicating local antigen‐driven selection [45]. Recent single‐cell and spatial studies have refined this view and suggest that extrafollicular differentiation is a major, and likely predominant, local B‐cell program in LN. Single‐cell RNA‐seq first identified local B‐cell activation associated with an ABC signature in LN kidneys [48, 49, 50]. This was later extended by studies showing enrichment of T‐bet+ ABCs clonally linked to adjacent PBs/PCs, providing direct evidence of in situ EF differentiation and local humoral activation [51]. Interestingly, GZMK+ CD8 T cells were linked to the extrafollicular B‐cell program, and ABCs were largely colocalized with these activated cytotoxic‐like T cells within renal lymphoid aggregates, suggesting that local T‐cell–derived inflammatory signals may help drive‐to‐PB differentiation. At the same time, lymphoid organization in LN should not be considered uniformly pathogenic in the same way. Spatial quantification studies showed that B‐cell–rich periglomerular neighborhoods, often associated with T follicular helper cells, can associate with better renal outcome, whereas small cytotoxic T‐cell neighborhoods in the tubulointerstitium predict progression to end‐stage renal disease [52]. This heterogeneity suggests that the key issue is not simply the presence or absence of lymphoid organization, but rather which cellular program predominates within that niche. In LN, current evidence argues that the most frequent and biologically relevant state is a tubulointerstitial ILS/mTLS niche in which B cells, T follicular helper‐like helpers, GZMK+ CD8 T cells, IFN/TLR signals, and inflammatory stromal cues cooperate to sustain local extrafollicular activation, PB differentiation, and probably organ‐focused repertoire diversification. Finally, spatial studies further indicate that LN kidneys contain specialized inflammatory niches enriched in stromal and myeloid programs that may sustain these responses. In particular, VCAM1+ proximal tubular cells were identified as components of an LN‐specific inflammatory/failed‐repair niche, where they interact with myofibroblasts and immune cells and may help maintain a pro‐inflammatory microenvironment permissive for ongoing lymphocyte activation [53]. In parallel, a large human–mouse atlas of LN myeloid cells identified two functionally distinct disease‐associated myeloid populations: a classical type 2 (C2) monocytes/macrophages cluster (CD9+, TREM2+, CD63+) associated with activity/proliferative lesions, and a resident macrophages cluster (RM0) (CD68+, CXCR1+ CD81+) linked to chronicity. These findings reinforce the concept that renal stromal and myeloid compartments are not passive bystanders but active organizers of local immune states [54]. However, the extent to which stromal and myeloid inflammatory niches directly shape local B‐cell activation and extrafollicular differentiation in LN remains unclear and warrants further investigation. IgA Nephropathy In IgA nephropathy (IgAN), chronic mesangial IgA deposition and complement activation generate a persistent inflammatory microenvironment that extends into the tubulointerstitium in the kidney [55]. Within this context, LS can develop in renal parenchyma, reflecting structured local immune activation rather than diffuse inflammation. Across studies, LS have been reported in 37.5%–50.8% of IgAN biopsies [56, 57]. These lesions are mainly composed of T cells and B cells consistent with ILS, whereas only a minority acquire features of more advanced organization, FDCs and proliferating GC‐like B‐cell clusters, consistent with mTLS. In a prospective cohort of 845 patients with IgAN, interstitial LS were present in 50.8% of diagnostic biopsies. Among LS‐positive cases, 17.1.8% displayed mTLS [56]. ILS were predominantly located in subcapsular, perivascular, and periglomerular regions and correlated with the severity and arterial lesions [56]. Clinically, the presence of both ILS and mTLS were associated with poor response to immunosuppressive therapy suggesting that in IgAN, LS may reflect a compartmentalized, self‐sustaining intrarenal immune response that is relatively refractory to conventional corticosteroid‐ or mycophenolate‐based strategies [57]. The precise trigger initiating ILS formation in IgA nephropathy remains uncertain. While persistent immune complex deposition and complement activation provide a plausible inflammatory driver, it is not clear whether its development is strictly dependent on ongoing local antigenic stimulation or can arise as a consequence of sustained tissue inflammation in the absence of a clearly identifiable antigen. Recent transcriptomic analyses suggest that renal lesion‐associated genes in IgAN are enriched in cytokine–cytokine receptor interaction pathways and correlate with increased infiltration of B cells, activated helper T cells, PC, and macrophages, supporting a model in which chronic inflammatory signaling and immune cell recruitment may be sufficient to promote lymphoid organization [58, 59]. Thus, whether intrarenal ILS in IgAN are directly immune complex–driven or instead represent a secondary, partially antigen‐independent process of chronic inflammatory structuration remains an open question. ANCA‐Associated Vasculitis In ANCA‐associated necrotizing crescentic glomerulonephritis, tubulointerstitial inflammation frequently includes organized LS that span a spectrum of organization. Early work by Steinmetz et al. [44] established a morphologic classification of intrarenal cell infiltrates ranging from scattered lymphocytes to nodular aggregates with compartmentalized B‐ and T‐cell zones. Within this continuum, scattered and clustered infiltrates correspond to inflammatory lymphoid structures, whereas nodular, compartmentalized aggregates represent advanced LS approaching mature TLS organization. Building on this framework, Brix et al. [60] prospectively analyzed 112 patients with newly diagnosed ANCA‐associated glomerulonephritis and confirmed the presence of four distinct grades of lymphoid organization using CD3/CD20/CD21 staining. Nearly half of biopsies contained clustered or nodular aggregates, and 15 (13.3%) cases displayed compartmentalized B‐ and T‐cell zones, with central CD21+ follicular dendritic cells networks. Importantly, the degree of lymphoid organization was strongly associated with renal outcome. Patients with clustered or nodular LS had significantly worse renal function at baseline and during follow‐up, and Kaplan–Meier analysis demonstrated markedly reduced renal survival compared with patients displaying only scattered infiltrates. Notably, lymphoid organization did not correlate with interstitial fibrosis, tubular atrophy, glomerulosclerosis, or crescent percentage, indicating that LS maturation reflects active immune structuration rather than simply chronic structural damage. In granulomatous lesions from granulomatosis with polyangiitis (GPA) (mostly associated with anti‐Proteinase 3 antibodies), mTLS and ILS were identified in close proximity to PR3+ B cells and PCs, and immunoglobulin gene analyses revealed extensive somatic hypermutation, clonal diversification, and patterns consistent with antigen‐driven selection and memory B‐cell differentiation [61, 62]. In particular, VH genes derived from these lesions showed structural similarities to PR3‐ANCA–encoding sequences and, especially in active disease, mutations predicted to favor binding to the positively charged PR3 autoantigen, suggesting that selection of potentially PR3‐reactive B cells may begin within granulomatous tissue [61]. Single‐cell analysis further confirmed clonal expansion and affinity maturation of tissue B cells within PR3‐associated granuloma [62]. Together, these studies support the idea that granulomatous inflammatory lesions in ANCA‐associated vasculitis not merely sites of nonspecific inflammation, but can function as local niches for autoreactive B‐cell selection and possibly early PR3‐ANCA generation. Transplantation In kidney transplantation, LS develop within the allograft as a consequence of sustained alloantigen exposure and chronic tissue injury, forming a continuum that ranges from diffuse ILS to highly organized posttransplant tertiary lymphoid organs or mTLS [63]. Early descriptions of rejecting grafts identified PNAd+ high endothelial venules and dense lymphocytic infiltrates, suggesting that the graft itself may acquire lymph node–like properties. Subsequent studies demonstrated clear T‐ and B‐cell compartmentalization, FDC networks, GC–like reactions, lymphangiogenesis, and local expression of lymphoid chemokines such as CXCL13, CCL19, and CCL21. These findings established that chronic rejection can recapitulate key elements of secondary lymphoid organogenesis within the transplanted organ, including local antigen presentation and B‐cell maturation. Mechanistically, lymphotoxin‐αβ/LTβR signaling, stromal activation, and sustained chemokine gradients appear central to posttransplant tertiary lymphoid organs induction and maintenance, although their requirement is context‐dependent and may vary according to the inflammatory milieu [64]. Clinically, most human and experimental data associate organized posttransplant tertiary lymphoid organs with chronic rejection and inferior graft survival, supporting the concept that the graft can become a site of autonomous alloimmune amplification. In particular, intragraft B‐cell clusters have been shown to exhibit clonal expansion, somatic hypermutation, and limited overlap with peripheral repertoires, consistent with local humoral immune responses [65, 66]. However, the functional significance of these structures is not uniformly pathogenic. As discussed in recent syntheses, posttransplant tertiary lymphoid organs display considerable plasticity: while frequently linked to chronic rejection, they may also, under certain conditions, harbor regulatory T‐cell–enriched microenvironments associated with graft acceptance in experimental models [67]. Thus, rather than representing a binary marker of rejection, intragraft LS likely reflect a dynamic spectrum of immune organization shaped by the balance between effector and regulatory pathways. Immune‐Related Adverse Events Immune checkpoint inhibitors (ICIs) targeting cytotoxic T‐lymphocyte–associated protein 4 (CTLA‐4) and the programmed cell death protein 1 (PD‐1) pathway, including its ligands programmed death‐ligand 1 (PD‐L1) and programmed death‐ligand 2 (PD‐L2), are widely used across many cancer types but can induce a spectrum of immune‐related adverse effects (irAEs). By blocking immune checkpoints that maintain peripheral tolerance and restrain autoreactive responses, these therapies can lead to autoimmune‐like manifestations affecting multiple organs [68]. Depending on the regimen used, 66%–96% of patient's experience treatment‐related toxicities from ICIs, and 7%–59% develop severe irAEs that could be ultimately life‐threatening [69]. These toxicities can involve virtually any organ, including the lungs, heart, joints, thyroid, pituitary, liver, colon, nervous system, and skin. Although most studies of irAE lesions primarily describe T‐cell infiltration, a few have reported more organized LS. In preclinical models, PD‐(L)1 blockade in aged tumor‐bearing mice induces multiorgan toxicities associated with CD4+ T cell–derived interleukin‐21 and increased expression of CXCL13, promoting the recruitment of both B and T cells resembling LS. These toxicities appear to be mediated by immunoglobulin G (IgG) deposition, and corticosteroid treatment reduces the accumulation of these lymphoid aggregates [70]. Consistent with these findings, by studying dermatologic irAEs, including ICI‐induced lichen planus, we identified immune infiltrates distinct from those observed in classical, non–drug‐induced lichen planus. In particular, ICI‐induced lichen planus' lesions exhibited reduced frequencies of CD4+ T cells and increased infiltration by B cells and macrophages, with immune cells organized into B‐cell aggregates surrounded by CD4+ and CD8+ T cells, suggesting LS organization [71]. Together, these observations support a potential role for these lesions in amplifying local autoimmune responses during checkpoint blockade. However, the nature of these LS remains to be better defined (Figure 2). Their prevalence, degree of organization, cellular composition, and spatial organization across affected organs have not yet been systematically quantified or compared. In particular, it remains unclear whether these aggregates mostly correspond to LS. Comparative analyses across different irAEs, and relative to ILS and mTLS found in autoimmunity or cancer, will be required to determine whether they represent a shared pathogenic program or tissue‐specific forms of local immune organization (Figure 2). Presence of lymphoid structures in minor salivary glands across different inflammatory contexts. One section of FFPE blocks of minor salivary glands from different inflammatory conditions were cut and processed by multiplex immunofluorescence. Images highlight the presence of LS in healthy aged individual (Sicca Syndrome), in a specific autoimmune disease (Seropositive Sjögren disease) characterized by the presence of specific autoantibodies in the serum and, in IRAEs without any autoantibodies reported (ICI‐induced Sicca Syndrome). Aging Aging is associated with profound alterations in immune function collectively referred to as immunosenescence. One of its main features is the development of chronic, low‐grade systemic inflammation, a process termed inflammaging which contributes to increased susceptibility to age‐related diseases [72]. In parallel, LS have been increasingly reported in several tissues of aged individuals and aged mice. In the lung, a tissue continuously exposed to environmental stress, aged mice (24 months old) exhibit a higher frequency of lymphoid aggregates compared with young mice (2 months old). These structures are predominantly located in perivascular, peribronchial, and parenchymal regions, with the highest density observed around blood vessels. They appear more compact and organized than LS induced by acute injury such as bleomycin treatment. These aggregates contain both T and B lymphocytes in central areas and macrophages, distributed more peripherally [73]. Age‐associated LS have also been described in salivary glands. In mice, immune cell infiltration begins around 8 months of age and progressively increases in density between 13 and 26 months, whereas young animals (3 months) show little to no infiltration. This process is accompanied by increased expression of chemokines involved in lymphocyte trafficking. With increasing age and size of the LS, the organization of B and T cells becomes more pronounced, resembling structures observed in autoimmune conditions such as SjD (see Section 2.2). Interestingly, ABCs were enriched in the salivary glands of old mice compared with other organs such as the spleen or liver [74]. Similar LS can also be observed in salivary glands from healthy elderly human donors (Figure 2). In the kidney, age‐dependent immune clustering has also been reported. A broad analysis across more than 20 mouse strains revealed clusters of CD45+ immune cells, predominantly B and T lymphocytes, whose cumulative size increased progressively with age in both males and females [75]. Supporting these findings, transcriptomic analyses of normal human kidney samples revealed increased expression of immune‐related genes with aging, including immunoglobulin chains (μ, λ, κ) and the T‐cell receptor β chain [76]. Additional studies using nephrectomy samples from individuals without renal cell carcinoma have also reported lymphoid aggregates more frequently in elderly (> 60 years) than in younger individuals (< 40 years), with some lesions displaying Ki67+ proliferating cells and FDC networks [77]. Age‐associated LS have also been described in the bladder. In aged mice (18 months vs. 3 months), transcriptomic analyses reveal increased activation of B‐ and T‐cell–related pathways together with upregulation of CXCL13. These changes were associated with higher frequencies of B, mostly naïve, and T lymphocytes in aged bladder tissue. Interestingly, this process occurs independently of the microbiota, as similar findings were observed in germ‐free mice, but it involved TNF‐α signaling, since aged TNF‐deficient mice display reduced lymphocyte accumulation and fewer aggregates [78]. The liver also exhibits age‐related immune infiltration. Compared with young mice (3 months), aged mice (24 months) show increased infiltration by macrophages, T cells, B cells, natural killer cells, and neutrophils. These cells often form focal accumulations in perivascular and lobular regions, with a predominance in perivascular areas, suggesting enhanced immune cell extravasation with aging. This is accompanied by elevated expression of multiple inflammatory mediators, including IFN‐γ, IL‐1β, IL‐6, TNF‐α, and IL‐2, as well as regulatory cytokines such as TGF‐β and IL‐10. Increased levels of chemokines such as RANTES, MIP‐1α, IP‐10, and GRO‐1 further support the recruitment of immune cells to aged liver tissue [79]. Collectively, these observations indicate that aging promotes the progressive accumulation of LS across multiple organs, even in the absence of overt autoimmune disease. While the functional consequences of these structures remain incompletely understood, their presence may also exacerbate chronic inflammation and interfere with normal tissue homeostasis and repair. Functional Roles of ILS and mTLS in Disease Pathogenesis Rather than differing only in their degree of maturation/organization, ILS and mTLS could both contribute to disease progression through alternative modes of immune amplification within tissues [8]. Rather than being passive bystanders of inflammation, these structures can influence the intensity, quality, and persistence of tissue immunity. Their functional impact depends on their degree of organization, cytokine environment, and cellular composition, positioning them as potential drivers or regulators of disease pathogenesis. Local Amplification of Autoreactive Responses The pathogenic role of LS in inflamed tissues has been most convincingly demonstrated in highly organized, germinal center–like mTLS [5, 21]. In these settings, local B/T segregation, follicular dendritic cell networks, and GC features expression (Ki67, AID, Bcl6) enable somatic hypermutation, class‐switch recombination, and affinity‐based clonal selection, directly supporting autoreactive repertoire diversification. Laser capture microdissection and immunoglobulin repertoire sequencing have shown extensive somatic hypermutation (SHM), intraclonal diversification, and lineage relationships between mTLS B cells and adjacent PB and PC in RA synovium and SjD salivary glands [23, 80, 81]. Within these mTLS, persistent antigen availability is sustained by FDC‐retained immune complexes and exposure to tissue‐specific autoantigens, including citrullinated proteins in RA and Ro/SSA–La/SSB ribonucleoproteins in SjD. Unlike SLO, mTLS develop in tissues lacking predefined lymphatic architecture and may exhibit impaired antigen drainage, favoring sustained local stimulation. The inflammatory milieu further reinforces this process. High levels of type I and II interferons, BAFF, IL‐7, IL‐21, IL‐17, and IL‐22 lower B‐cell activation thresholds, promote T follicular helper differentiation, and support both GC‐like and extrafollicular differentiation programs. In SLE, convergent evidence supports a similar paradigm of local autoreactive amplification within inflamed tissues. Immune aggregates resembling mTLS with true GC‐like features have been observed in the kidneys of patients with lupus nephritis (see Section 2.3), but with low frequencies. In lupus‐prone NZB × NZW F1 mice, transcriptomic profiling revealed a LN‐like TLS gene signature closely mirroring that of SLO [82]. Renal mTLS in lupus are associated with heightened local autoimmunity and disease severity, correlating with chronic tubulointerstitial inflammation and poor response to therapy, supporting a pathogenic role for in situ B‐cell activation and autoantibody production [83]. Importantly, despite their GC‐like architecture, mTLS may not fully reproduce the regulatory stringency of bona fide secondary GCs. Under physiological conditions, functional GCs actively counterselect autoreactive clones. Elegant work by Singh et al. [84] showed that pathogenic autoantibodies can emerge only after acquisition of lymphoma‐associated signaling mutations (CARD11, TNFAIP3, KLHL6, or BTG2), which precede pathogenic V(D)J diversification, suggesting that altered survival signaling is required to bypass normal GC selection constraints. Notably, recent observations suggested that AID expression within mTLS was often quantitatively reduced or spatially dysregulated compared with classical GCs, potentially decreasing selection stringency and allowing survival of autoreactive clones that would otherwise be eliminated [85]. Additionally, the alterations observed in LS may be amplified from dysregulation of T‐cell–mediated control mechanisms that normally enforce GC tolerance. Follicular regulatory T cells directly restrain autoreactive B‐cell responses and are required to prevent antinuclear antibody emergence [86]. Imbalance in the T follicular helper/regulatory axis within inflamed tissues could therefore weaken local tolerance checkpoints, rendering mTLS more permissive to autoreactive B‐cell survival and diversification despite their GC‐like organization. By contrast, the functional contribution of ILS remains less clearly defined. Unlike mTLS, ILS lack organized GC and clear evidence of structured clonal selection. Their pathogenic impact may therefore arise less from affinity maturation than from their ability to establish permissive inflammatory microenvironments that promote immune cell retention and lower activation thresholds for autoreactive or inflammation‐adapted subsets. Rather than representing incomplete mTLS, ILS may correspond to inflammation‐dominant states in which antigen engagement is limited or not yet structurally organized. Antigen persistence may subsequently stabilize these aggregates and drive their maturation toward mTLS. Distinguishing these states is therefore essential to determine whether lymphoid organization primarily amplifies inflammation, supports clonal diversification, or integrates both processes over time. Microenvironmental Reprogramming and Cellular Specialization: Survival Niches/Plasma Cells Beyond their role as sites of local antigen‐driven diversification, LS may function as survival niches that sustain long‐lived immune populations within inflamed tissues. In physiological settings, PC survival is typically confined to specialized bone marrow niches providing APRIL, BAFF, IL‐6, CXCL12, and stromal support. However, chronically inflamed tissues can recapitulate key elements of this microenvironment. Stromal cells, myeloid populations, and activated T cells within LS produce survival factors that support PB differentiation and PC persistence in situ. In mTLS, survival niche formation may directly be coupled to ongoing differentiation as GC‐like–competent structures generate PB locally while simultaneously providing the stromal and cytokine support required for their stabilization. Interestingly, this observation has been well established in mTLS from cancer [87, 88] whereas its formal demonstration in autoimmune LS remains more limited. In parallel, accumulating evidence suggests that memory B cells can establish tissue residency programs characterized by altered chemokine receptor expression, reduced egress signals, and adaptation to local inflammatory cues [89]. In this context, ILS may be particularly well suited to imprint inflammatory specialization rather than classical affinity maturation. These structures arise in cytokine‐rich, often IFN‐dominated environments that favor the expansion of inflammation‐responsive populations. Such conditions may promote the local establishment of resident memory B cells as well as the accumulation of EF Tbet+ ABCs that are particularly IFN‐responsive. Rather than refining the repertoire through affinity maturation, ILS may therefore influence its functional polarization by favoring subsets with heightened responsiveness to innate and inflammatory stimuli. Similarly, chronically activated CD4+ T cells, including T peripheral helper–like populations, may persist within ILS and provide B‐cell help independently of classical follicular organization. In this model, ILS function less as sites of clonal evolution than as microenvironments that imprint inflammatory specialization, tissue adaptation, and local persistence. Such a framework may be particularly relevant in diseases where ILS predominate over mTLS, including lupus nephritis, SjD, or IgAN, and where local immune activation appears strongly linked to EF programs. Regulatory and Protectives Niches LS can support distinct functional outcomes, as their local impact depends not only on the architecture but also on the immune states they possess. In tumors, numerous studies indicate that mTLS, enriched in follicular B cells, GC‐like B cells, PB/PC, and memory B cells, are usually associated with productive local immunity, including coordinated T/B cell interactions, in situ antibody generation, and improved clinical outcome [90, 91, 92]. This protective configuration has been repeatedly linked to the presence of mature GC‐containing TLS and to local tumor‐reactive humoral responses, suggesting that fully organized mTLS can function as privileged niches of effective antitumor immunity. However, LS quality is heterogeneous. In melanoma and other tumors, heterogeneity in maturation state, cellular composition, and spatial organization is associated with distinct immune outputs and clinical trajectories, indicating that not all LS are functionally equivalent [93, 94]. A central variable in this functional divergence seems to be the B‐cell program established within the niche. Recent reviews emphasize the dual immunomodulatory role of B cells in solid tumors: when organized within mTLS, B cells generally contribute to antitumor immunity; by contrast, in less structured ILS or more suppressive environments, Breg‐like populations may inhibit effector lymphocytes through IL‐10, TGF‐β, IL‐35, adenosine, or contact‐dependent suppressive pathways [91, 95, 96]. Tumor infiltrating B cells have been recently extensively characterized using a broad range of approaches, including single‐cell RNA sequencing [97] Phenotypically, they span the full continuum from naive B cells to PCs, but are often enriched in memory B cells, PB, and PCs. In some cancers, B cells with an ABC‐like phenotype have also been described. These populations inversely correlate with affinity‐matured B cells in lung cancer [98], and, unlike mTLS, are enriched in nonresponders to checkpoint blockade in melanoma [2]. In this context, mTLS are generally interpreted as GC‐like structures, supporting organized T–B collaboration, affinity maturation, and memory/PC generation and potentially antitumor immunity. By contrast, ILS may correspond more closely to EF‐like inflammatory niches, in which B‐cell responses are less structured and may instead favor ABC‐like populations and potentially suppressive B‐cell states. Consistent with this, in NSCLC, tumors rich in B cells show higher proportions of naive, central‐memory, and activated CD4+ T cells, but lower frequencies of immune checkpoint–expressing CD4+ T cells and Tregs, and these TLS‐Bhigh/Treglow configurations identify the best clinical outcomes [99]. However, the opposite configuration can also emerge [100]. The clearest recent example is the identification of IGLL5+ B cells in bladder cancer. Rather than acting through classical Breg cytokines alone, these cells suppress immunity by destabilizing LS structure itself: they interact with LTβR on HEVs, inhibit noncanonical NF‐κB signaling, induce HEV dysfunction, impair lymphocyte recruitment, and ultimately drive LS disassembly, leading to weaker CD8+ effector activation and poorer response to checkpoint blockade. This finding shows that B cells can make LS suppressive not only by producing inhibitory mediators, but also by disrupting the vascular and stromal infrastructure required for protective LS homeostasis. It also resonates with the idea that location matters. mTLS positioned near antigenic sources tend to support protective responses, whereas distally organized LS may instead sustain inflammation, or even contribute to immunosuppressive responses. This interpretation is consistent with studies of immature B‐cell programs in tumors: in ccRCC, IRF4high immature B‐cell states appear linked to early LS initiation but remain associated with an inflammatory state (referred to as TLS immaturity), and disease progression, suggesting that persistent immature B‐cell recruitment may restrain full TLS maturation and thereby limit effective antitumor responses [94]. Thus, some ILS may represent niches of partial organization in which B‐cell recruitment occurs, but maturation may be arrested, diverted, or coupled to inhibitory pathways. Taken together, LS can support at least three nonexclusive “protective/regulatory” outputs: (1) mature protective TLS that organize effective local immunity; (2) checkpoint‐restrained but still tissue‐protective niches limiting excessive damage; and (3) B‐cell–driven suppressive states that become maladaptive in cancer by undermining antitumor responses. By contrast, in autoimmunity, the regulatory or tissue‐protective functions of LS remain far less well defined and have been comparatively little explored. Yet, this imbalance in the literature should not exclude the possibility that, under specific conditions, LS in autoimmune tissues may also acquire regulatory features. In this context, the key issue is not whether LS are intrinsically beneficial or harmful, but which immune cell states, stromal programs, and cytokine milieus they stabilize locally. This raises the possibility that the microenvironment may actively reprogram LS functional output. Such a model is reinforced by our recent work showing that inflammatory‐prone ABC precursors can be redirected by IL‐2 toward a MAF‐dependent IL‐10+ regulatory state, suggesting that even B‐cell programs classically linked to chronic inflammation may remain plastic and context‐dependent [101]. This possibility may be particularly relevant for ILS, where local cues could therefore determine whether ABC‐like states reinforce pathology or instead acquire regulatory function. Therapeutic Targeting of TLS In cancer, inducing mTLS and ILS to optimize their function has emerged as a promising immunotherapeutic strategy, particularly to enhance the efficacy of immune checkpoint inhibitors [7]. Several approaches have been developed in preclinical models to promote LS formation or enhance their activity within the tumor microenvironment [7]. These strategies include the delivery of lymphoid chemokines such as CCL19, CCL21, and CXCL13; the use of antibody–lymphotoxin‐α fusion proteins; and immune‐stimulating agents, including agonists of the stimulator of interferon genes (STING) pathway and CD40 agonistic antibodies. Additional strategies rely on stromal cells capable of organizing lymphoid niches, antibodies targeting regulatory immune cells, oncolytic viruses engineered to promote immune activation, and chimeric antigen receptor T cells designed to remodel the tumor microenvironment. By contrast, in autoimmune diseases, LS are generally associated with disease progression, suggesting that interfering with their formation or maintenance could be therapeutically beneficial. Accordingly, several studies have evaluated the impact of immunomodulatory treatments on LS organization and activity in autoimmune tissues. CD20‐Targeted Therapy B‐cell depletion therapy represents one approach capable of affecting LS. Rituximab, a chimeric monoclonal antibody directed against CD20, has been evaluated in SjD patients. In the TEARS study, analysis of minor salivary gland biopsies demonstrated that clinical responders to rituximab exhibited a significant reduction in the focus score [102]. Similarly, in a double‐blind, placebo‐controlled trial in SjD, sequential parotid gland biopsies obtained at baseline and 12 weeks after treatment showed that rituximab therapy resulted in a marked reduction in B‐cell infiltration and decreased density of GC–like structures within the glands [103]. However, findings from other disease settings suggest that LS may also confer resistance to systemic B‐cell depletion. In chronic renal allograft rejection, rituximab treatment failed to induce LS regression in the kidney despite effective depletion of peripheral B cells [104], indicating that local LS microenvironments may support B‐cell persistence and limit the impact of systemic CD20‐targeted therapy. BAFF‐Axis‐Targeted Therapy Targeting the B‐cell activating factor (BAFF) pathway represents another strategy to interfere with TLS/ILS‐associated B cell response. BAFF is a key survival and differentiation cytokine for B cells and plays a central role in the pathophysiology of autoimmune diseases. In SjD, increased expression of BAFF and its receptors has been reported in minor salivary glands of patients displaying mTLS [105]. These observations support the hypothesis that BAFF–BAFF receptor interactions contribute to the establishment and maintenance of these niches and may drive disease progression in a subset of patients. Therapeutic blockade of BAFF has demonstrated clinical benefit in SLE [106]. Belimumab, a monoclonal antibody targeting soluble BAFF, is approved for SLE and is associated with reduced disease activity and flare rates. Other agents targeting this pathway, including tabalumab (anti‐BAFF monoclonal antibody), blisibimod (BAFF‐neutralizing peptibody), atacicept (TACI–IgG Fc fusion protein), and more recently ianalumab (anti‐BAFF receptor monoclonal antibody), have also been evaluated in clinical trials [107, 108], further highlighting the therapeutic relevance of BAFF signaling. Experimental evidence supports a more direct link with LS biology. In murine lupus models, increased BAFF production has been linked to the induction of renal LS and exacerbation of lupus nephritis [109]. Conversely, in vivo BAFF reduction prevented LS formation and improved renal outcomes, although residual immune infiltrates and immune complex deposition persisted. These findings suggest that BAFF contributes not only to LS formation but also to their spatial organization, particularly by influencing T‐cell positioning within inflamed tissues. Collectively, both preclinical and clinical data indicate that BAFF‐targeted therapies attenuate autoreactive B‐cell responses and reduce disease activity, supporting the concept that modulation of the BAFF axis may represent an effective strategy to disrupt TLS formation or function in autoimmune diseases. CXCL13‐Targeted Therapy Targeting chemokines involved in lymphoid follicle formation has been explored in preclinical models. In a mouse model of SLE with neurobehavioral abnormalities reminiscent of human neuropsychiatric SLE, female MRL/lpr mice were treated with a neutralizing antibody against CXCL13, a key chemokine involved in lymphoid follicle formation. Anti‐CXCL13 treatment attenuated behavioral deficits in lupus‐prone mice. However, CXCL13 blockade did not substantially alter the extent of lymphocyte infiltration into the brain and the overall composition of the immune aggregates [110]. T Cell‐Targeted Therapy Targeting the inducible T‐cell costimulatory (ICOS) pathway represents a potential strategy to modulate LS. ICOS signaling promotes differentiation of Tfh cells and GC formation, and expansion of ICOS+ Tfh‐like and T peripheral helper–like cells has been associated with mTLS in pSS [111]. Increased PD‐1+ICOS+ T‐cell infiltration and upregulation of the ICOS–ICOS ligand axis in affected salivary glands further support its role in sustaining local inflammatory and B‐cell–driven immune responses. Experimental blockade of ICOS in ex vivo salivary gland organ cultures reduces pro‐inflammatory cytokine production, highlighting its functional relevance. However, clinical targeting with the anti‐ICOSL antibody MEDI5872, although associated with reductions in ICOS+ Tfh‐like cells in minor salivary glands and in rheumatoid factor levels, has not demonstrated consistent clinical efficacy. Collectively, these findings indicate that the ICOS pathway contributes to LS‐associated immune activation but may require combination approaches to achieve meaningful therapeutic benefit. Modulation of interleukin‐27 (IL‐27) signaling represents another potential therapeutic approach to limit LS development through regulation of CD4+ T cells. Reduced IL‐27 signaling in humans and mice promotes LS formation, which is associated with severe disease and resistance to biological therapy in RA [112]. Increased numbers of podoplanin‐expressing Th17 cells in the absence of IL‐27R signaling may drive LS formation. Strategies that restore or enhance IL‐27 signaling could therefore disrupt the cellular and cytokine networks sustaining LS formation and function. Overall, these findings highlight the potential of targeting LS as a therapeutic strategy in autoimmune diseases. Future efforts should focus on developing novel approaches to suppress LS formation and function, as well as conducting more mechanistic studies to elucidate the cellular and molecular pathways that sustain these structures. Technologies Redefining ILS and Mature TLS Detection There is little doubt that, in the coming years, emerging technologies will substantially refine our understanding of LS, including their development (antigen‐driven vs. inflammation‐driven), their cellular composition, their functional outputs (protective vs. regulatory), and ultimately the most effective strategies to target them (promotion vs. inhibition). Among these technologies, the integration of spatial transcriptomics (ST) with artificial intelligence (AI)–based machine learning holds particular promise for advancing LS research. To illustrate how ST could nuance our understanding of LS, we analyzed an infiltrated salivary gland from a SjD patient. Immunohistochemistry (IHC) revealed numerous LS within the analyzed tissue (Figure 3A). When mapping signatures derived from Nayar et al., incorporating both SLOs and inflammation‐associated genes, LS could be appreciated with a more nuanced view (Figure 3B). Indeed, while some LS could be associated with mTLS (red square), displaying CXCL13 expression, Tfh and PC presence (Figure 3C,D), other LS displayed a higher CXCL9 and CXCL10 expression, a decreased presence of Tfh and PC signature (blue square). These observations highlight that ST could uncover marked heterogeneity among LS. Spatial transcriptomics guidance in distinguising the different structures. Serial sections of one FFPE block of a minor salivary gland of a SjD patient were cut and processed by IHC multiplex and spatial transcriptomics using the Visium HD platform of 10× genomics [113]. (A) Visualization of section of the minor salivary glands stained with CD20—pink, CD3—cyan, CD68—yellow, CD138—green. (B–E) Visualization of transcriptomic signatures on the serial section of the same minor salivary gland extracted from (B) Nayar's paper which identified SLOs and inflammatory signatures [40], and (C, D) Fridman's review [114] which highlighted signatures for (C) migration ability, and (D) specific cell types. Looking ahead, once such LS states are better defined, AI tools could be trained on ST datasets of LS and be used to predict the type of LS directly from standard H&E sections. Recent developments have already proven the feasibility of such approaches. Deep‐learning approaches can now predict single‐cell gene expression from H&E images, extending earlier spot‐level prediction frameworks toward finer‐resolution molecular inference [115]. Other studies have shown that AI can generate virtual immunohistochemistry from hematoxylin and eosin slides, offering a way to infer tissue‐marker patterns without additional staining [116]. In parallel, graph‐based models combined with ST have been used to improve routine histologic and molecular classification, illustrating how molecularly informed AI may support diagnostic interpretation predicting histology and molecular analysis from ST data [117]. Together, these developments suggest that ST‐AI integration provides a powerful framework to resolve ILS heterogeneity, classify LS along inflammatory‐to‐organized continua, and ultimately guide more precise therapeutic strategies targeting these structures. Conclusions Taken together, the studies discussed here support an alternate view of LS in inflamed tissues, describing ILS which can be considered as immature TLS, but also as potentially distinct inflammatory states shaped by the balance between antigen exposure, stromal organization, and local tolerance control. Inflammation may provide the permissive conditions for lymphoid assembly, whereas persistent antigenic stimulation and defective regulatory mechanisms may drive structural stabilization, clonal evolution, and, in some settings, full TLS maturation. This model may be particularly relevant in aging, where partially organized lymphoid states can emerge in tissues even in the absence of a clearly defined antigenic trigger. Altogether, this perspective suggests that LS may be better viewed as a heterogeneous set of tissue lymphoid niches rather than through a binary classification, with functional outputs that are highly context‐dependent. Depending on the tissue environment, these structures may promote protective immunity, pathogenic autoreactivity, or regulatory adaptation. Several questions therefore remain open: Are these LS necessarily antigen‐driven structures, or can they arise primarily from an inflammatory environment?Does the absence of germinal center formation reflect incomplete maturation, or a distinct functional state?Can LS in autoimmune tissues also support protective or regulatory niches?To what extent can LS be functionally reprogrammed by local cytokine, stromal, or cellular cues?And what determines whether these structures promote protective immunity, pathogenic autoreactivity, or immune regulation? Are these LS necessarily antigen‐driven structures, or can they arise primarily from an inflammatory environment? Are these LS necessarily antigen‐driven structures, or can they arise primarily from an inflammatory environment? Does the absence of germinal center formation reflect incomplete maturation, or a distinct functional state? Does the absence of germinal center formation reflect incomplete maturation, or a distinct functional state? Can LS in autoimmune tissues also support protective or regulatory niches? Can LS in autoimmune tissues also support protective or regulatory niches? To what extent can LS be functionally reprogrammed by local cytokine, stromal, or cellular cues? To what extent can LS be functionally reprogrammed by local cytokine, stromal, or cellular cues? And what determines whether these structures promote protective immunity, pathogenic autoreactivity, or immune regulation? And what determines whether these structures promote protective immunity, pathogenic autoreactivity, or immune regulation? Resolving these questions will be critical to better define the biology and functional diversity of LS, and may ultimately open new therapeutic avenues to prevent their formation, limit their pathogenic activity, or reprogram them toward protective or regulatory states. Funding This study was carried out with the support of the European Union, the French Government, Région Bretagne, Conseil Départemental du Finistère, and Brest Métropole, within the framework of the CPER. Ethics Statement The study was performed in accordance with the Declaration of Helsinki and was approved by ethical committees. The Diagnostic Suspicion of Primitive Syndrome Sjogren's (DIApSS) study is registered on (https://clinicaltrials.gov/study/NCT03681964 | ClinicalTrials.gov) (identifier: NCT03681964). The use of iRAES patients samples was approved by the Brest Hospital ethics committee (B2022CE.35). Conflicts of Interest The authors declare no conflicts of interest.