Early mycosis fungoides: Integrating clinical, histopathological and molecular insights to improve diagnosis and risk stratification

Abstract

Mycosis fungoides (MF) remains a diagnostic challenge in its early stages, due to its ability to mimic common inflammatory dermatoses and the subtlety of its histopathological features. This review integrates recent advances in anatomical pathology and molecular research, highlighting how the combined assessment of morphology, immunophenotype, clonality studies and emerging high-throughput platforms is reshaping our understanding of the disease. Transcriptomic analyses, microRNA profiling and spatial technologies reveal distinctive signatures of early-stage MF, along with microenvironmental and immune alterations that precede clinically evident progression. These findings not only enhance diagnostic accuracy but also help identify patients at higher risk of disease progression. Overall, the integration of clinical, histopathological, and molecular data is defining a new paradigm for refined prognostic stratification and precision-medicine approaches in cutaneous T-cell lymphomas.

Resumen

La micosis fungoide (MF) sigue siendo un desafío diagnóstico en sus fases iniciales debido a su capacidad para imitar dermatosis inflamatorias comunes y a la sutileza de sus características histopatológicas. Esta revisión integra los avances más recientes en anatomía patológica y en investigación molecular, destacando cómo la combinación de la morfología, el inmunofenotipo, los estudios de clonalidad y las tecnologías emergentes de alto rendimiento está redefiniendo nuestra comprensión de la enfermedad. Los análisis transcriptómicos, el análisis de microARN y las tecnologías espaciales revelan firmas distintivas de la MF en estadios precoces, junto con alteraciones del microambiente y del sistema inmunitario que preceden a la progresión clínica. Estos hallazgos no solo mejoran la exactitud del diagnóstico, sino que también permiten identificar a los pacientes con mayor riesgo de progresión de la enfermedad. En conjunto, la integración de datos clínicos, histopatológicos y moleculares está definiendo un nuevo paradigma para una estratificación pronóstica más precisa y para el desarrollo de medicina de precisión en los linfomas cutáneos de células T.

Introduction

Cutaneous T-cell lymphomas (CTCLs) are a heterogeneous group of extranodal non-Hodgkin lymphomas derived from skin-homing mature T cells. Although rare, with an incidence of 7–10 cases per million person-years, they represent the most common primary cutaneous lymphoma. Mycosis fungoides (MF) accounts for approximately 60% of CTCLs, followed by Sézary syndrome (SS) and primary cutaneous CD30-positive lymphoproliferative disorders (including lymphomatoid papulosis [LyP] and primary cutaneous anaplastic large-cell lymphoma [pcALCL]).^1–3

Early-stage MF generally follows an indolent course and typically presents as erythematous patches and plaques that may persist for years. Most patients remain confined to the skin, whereas 20–25% progress to advanced stages, which significantly affect quality of life and prognosis (median survival 13–56 months).^1,3 Clinical and histopathological overlap with benign inflammatory dermatoses (BIDs) often leads to delayed diagnosis, underscoring the importance of accurate differential diagnosis.¹ Distinguishing early-stage MF from other inflammatory conditions is therefore essential, as early recognition allows closer monitoring, tailored management, and improved long-term outcomes.

MF is staged as early-stage (IA–IIA; 78–93% of cases) and advanced-stage (≥IIB; 7–22%). Five-year overall survival declines from over 90% in stage IA/IB to 70–85% in stage IIA, 50–70% in stage IIB, and below 30% in stages III–IVB. The modified TNMB system of Bunn and Lamberg (1979) is used to stage MF, incorporating T for cutaneous lesions, N for lymph nodes, M for visceral involvement, and B for peripheral blood.^3–5

Recent research focuses on molecular and immunological biomarkers capable of predicting disease progression or aiding differential diagnosis. High-throughput approaches, including transcriptomics, multiplex immunohistochemistry, and spatial profiling, reveal gene expression patterns associated with disease evolution and microenvironment remodelling. These advances are paving the way for precision medicine strategies, in which clinical management can be guided by the molecular and immune landscape of early-stage MF.

Integration of clinical, pathological, and molecular data is becoming essential to refine diagnosis, assess prognosis, and personalise management. Understanding the interplay of genetic, immune, and microenvironmental factors driving progression remains a major challenge, and addressing this complexity is key to improving patient outcomes and preventing progression to advanced disease. In this context, this review aims to integrate current clinical, histopathological, immunophenotypic, and molecular insights relevant to the diagnosis and risk stratification of early-stage mycosis fungoides, with particular emphasis on established criteria, emerging biomarkers of disease progression, and their potential applicability in routine diagnostic practice.

Diagnosis of early-stage mycosis fungoides: histopathological, clinical, and molecular insights

Histopathological features and differential diagnosis

Histopathological examination of lesional skin is essential for the diagnosis of early-stage MF. The key pathological hallmarks include: (1) atypical lymphoid cells slightly larger than normal lymphocytes, with cerebriform, hyperchromatic nuclei; (2) lymphocytes distributed singly or in small clusters within an epidermis lacking spongiosis, a feature termed disproportionate epidermotropism; (3) isolated haloed atypical lymphocytes scattered throughout the epidermis; (4) alignment of individual atypical lymphocytes along the dermoepidermal junction; (5) papillary dermal fibrosis; and (6) a band-like lymphoid infiltrate in the dermis. Among these, the presence of atypical lymphoid cells within the epidermis (Fig. 1B) is often considered the most characteristic feature of early-stage MF. However, some cases show only minimal cytological atypia or limited epidermotropism. Moreover, epidermotropism-like patterns and subtle atypia may also occur in BIDs, making histological distinction from early-stage MF challenging in certain cases.^1,5–7

Fig. 1.

Histopathological and immunohistochemical findings in early-stage mycosis fungoides. (A) H&E of a classic patch showing epidermotropism without spongiosis. (B) Clonal TCRB expression in the epidermal and dermal infiltrate (10×). (C) Predominance of CD4⁺ T cells within the epidermis (20×). (D) Preservation of CD5 expression (10×). (E) Partial loss of CD7, typical of cutaneous T-cell lymphoma (10×). (F) CXCL13 positivity in a dermal subpopulation consistent with a TFH-like phenotype (10×). (G) PD-1 highlighting activated/TFH-like T cells in the epidermis (10×). (H) H&E of folliculotropic MF showing characteristic perifollicular infiltrates and follicular destruction.

In addition to these classic features, Dalton et al. noted that eosinophil infiltration exceeding three cells per section is uncommon in early-stage MF, suggesting that the degree of eosinophilia may aid in differentiating early-stage MF from BIDs. Nevertheless, because histopathology alone may be inconclusive, repeated biopsies or sampling from multiple lesions is often required to establish a reliable diagnosis.^1,6

Classical MF exhibits a characteristic immunophenotypic profile consistent with its derivation from mature, skin-homing T cells, typically expressing TCRβ⁺, TCRγ⁻, TCRδ⁻, CD3⁺, CD4⁺, CD5⁺, CD8⁻, and CD45RO⁺. Nonetheless, this profile may be heterogeneous, and immunophenotypic aberrancies can emerge during disease progression, highlighting the importance of integrating clinical, histopathological, and immunophenotypic findings for accurate diagnosis. Reported aberrant phenotypes include CD4⁻/CD8⁻ or CD4⁺/CD8⁺ combinations^1,5,8 (Fig. 1).

MF may present with immunophenotypic variants, including cases co-expressing follicular helper T-cell (TFH) markers.⁹ It may also express more than one TFH marker,¹⁰ making clinical correlation essential to exclude systemic TFH T-cell lymphomas with cutaneous involvement, which can closely mimic MF and primary cutaneous T-follicular helper lymphoma. Immunohistochemical studies have shown that, in some cases of MF, neoplastic cells express CD10, BCL6, PD-1, and occasionally CXCL13.¹⁰ Cases with a CD4⁺/CD8⁺ phenotype may be enriched for PD-1 positivity^9,11 (Fig. 1).

In addition, cases of MF with enrichment of regulatory T cells (Tregs) have been described. Tregs constitute a specialised immunosuppressive subpopulation characterised by high expression of CD25 and FOXP3, as well as CTLA-4 and GITR, and low expression of CD127. Infiltrating FOXP3⁺ Tregs are consistently detected in early-stage MF lesions, at levels comparable to those observed in BIDs, but decline with disease progression. Several studies suggest that a higher proportion of infiltrating Tregs is associated with a more favourable prognosis, likely due to suppression of neoplastic cell expansion.¹² Importantly, when a prominent FOXP3⁺ T-cell infiltrate is identified, particularly in cases with atypical clinical or histopathological features, adult T-cell leukaemia/lymphoma (ATLL) should be carefully excluded, as it may display a Treg-like phenotype with high FOXP3 expression and mimic MF clinically or histologically.¹²

MF may also exhibit a γδ phenotype, typically without clinical or prognostic differences.^13,14 Misclassification as primary cutaneous γδ T-cell lymphoma may result in unnecessary overtreatment. When MF expresses CD8⁺ or γδ profiles, clinical correlation remains essential to distinguish it from CD8⁺ epidermotropic T-cell lymphoma and primary cutaneous γδ T-cell lymphoma.¹³

MF encompasses several clinical variants, with distinct histopathological and immunophenotypic features. Among these, folliculotropic MF (FMF) is the most common clinicopathological subtype, accounting for about 12% of cases, and is defined by follicle-based lesions and predominant folliculotropism, with or without follicular mucinosis (Fig. 1H). Early FMF is typically associated with a favourable prognosis, whereas survival rates in tumour-stage FMF have been reported to be similar to those in classic tumour-stage MF. In most cases, neoplastic cells in FMF exhibit a CD3⁺/CD4⁺/CD8⁻ T-cell phenotype, similar to classic MF.^5,7 Other variants, including hypopigmented MF, poikilodermatous MF, pityriasis lichenoides-like MF, and localised pagetoid reticulosis (Woringer–Kolopp type), frequently show a CD8⁺ cytotoxic phenotype.^5,15,16 Recent single-cell transcriptomic data suggest that hypopigmented MF may not represent a true CD8⁺ neoplasm; instead, the malignant clone appears to be a small CD4⁺, undetectable by conventional immunohistochemistry and PCR-based clonality assays, while the dominant CD8⁺ population likely affects reactive expansion.¹⁷ The poikilodermatous variant often co-expresses CD8 and CD56,¹⁸ although CD56 expression has also been reported in non-poikilodermatous MF.¹⁹

Abnormal SOX10 staining, reflecting melanocyte loss, has been reported in approximately half of hypopigmented MF cases and may occur in both CD4⁺ and CD8⁺ subtypes.^17,20,21 Rare cases demonstrate an immunophenotypic shift from CD4⁺CD8⁻ MF to a CD4⁻CD8⁺ profile at transformation, while retaining clonal identity.^22,23 Similar shifts have also been described in recurrent cases without large-cell transformation.²⁴ Other rare variants of MF include syringotropic MF, hyperpigmented MF, ichtyosiform MF, MF palmaris et plantaris, psoriasiform MF, unilesional MF, urticarial MF, poikilodermatous MF, pigmented purpuric dermatosis (PPD)-like MF, verrucous MF, acanthosis nigricans-like MF (vegetating or papillomatous MF), bullous MF, pustular MF, interstitial MF, granulomatous slack skin and granulomatous MF.^4,5,7

Diagnosing early-stage MF can be challenging due to its clinical and histopathological overlap with BIDs, including eczematous dermatitis, psoriasis, lichen planus, pityriasis rubra pilaris, pityriasis rosea, and hypopigmented disorders such as vitiligo or pityriasis alba, as some MF cases closely mimic these conditions. Diagnosis therefore requires an integrated approach, combining clinical, histopathological, molecular, and immunopathological findings. To standardise this approach, the International Society for Cutaneous Lymphoma (ISCL) proposed a diagnostic algorithm in 2005 that assigns points to specific clinical features (persistent, progressive, non-sun-exposed patches, poikiloderma, and lesion variation), histopathological findings (superficial lymphoid infiltrate, epidermotropism, and lymphoid atypia), molecular evidence of T-cell clonality, and immunopathological abnormalities (T-cell antigen deficiencies or epidermal/dermal discordance). A total score of ≥4 supports the diagnosis of early-stage MF, enabling a flexible, integrative assessment in which no single criterion is essential, thereby improving reproducibility and facilitating multicentre standardisation⁶ (Fig. 2).

Fig. 2.

Integrated differential diagnosis criteria for early-stage mycosis fungoides versus benign inflammatory dermatoses (BIDs). The profiles of both conditions are compared across three levels of analysis: Histopathological (Step 1), Clinical (Step 2) and Molecular (Step 3). Integration of data from all three levels is essential to reach a definitive diagnosis.

Detection of T-cell receptor (TCR) gene monoclonality by PCR or Next-Generation Sequencing (NGS) represents another key diagnostic tool, with clonal TCR rearrangements identified in approximately 83% of early-stage MF cases. Clonality analysis most commonly targets the TCRγ locus due to its early rearrangement during T-cell development and technical advantages that facilitate PCR amplification. However, analysis of TCRβ rearrangements may increase diagnostic sensitivity, particularly in early lesions, and combined assessment of TCRγ and TCRβ maximises clonal detection while reducing false-negative results. Identification of identical T-cell clones in samples from different anatomical sites is highly specific and may provide important diagnostic support in challenging cases, although clonality assessment is not required in all patients^6,25,26 (Fig. 2). On the contrary, clonal peaks can be also found in reactive conditions.

TRBC1 immunostaining has recently emerged as a reproducible tool for assessing T-cell monotypia in MF and SS. As an immunohistochemical surrogate for TCRβ clonality, TRBC1, particularly when combined with CD3, can help distinguish neoplastic from reactive infiltrates. Proposed cutoffs of <25% or >75% TRBC1⁺ cells among CD3⁺ T cells may aid in determining monotypia.²⁷

The difficulty in differential diagnosis is compounded by the lack of tumour cell-specific markers. Classical markers, such as partial CD7 loss in <10% of T cells, support a diagnosis of MF but lack specificity.²⁸ More extensive or complete CD7 loss is uncommon in early lesions but may be more specific for neoplastic T cells. Conversely, loss of pan-T-cell markers, including CD2, CD3, or CD5, is more specific but rare in early lesions. Interestingly, partial CD7 loss or CD4/CD8 T double negativity may also be observed in reactive inflammatory conditions or following corticosteroids usage.²⁹

TOX has been proposed as a diagnostic adjunct, with expression increasing alongside disease progression. Early studies reported TOX expression in 74% of MF cases versus 32% in BIDs and normal skin; although not entirely tumour cell-specific, it may complement existing diagnostic algorithm. Other potential markers, such as TWIST1 and SATB1, have also been suggested in early-stage MF^1,30 (Fig. 2).

Clinical presentation and differential diagnosis

The initial presentation of MF is highly variable and nonspecific. The classic progression is from patches to plaques and eventually to tumours, but early disease is predominantly confined to the patch and plaque stages.^6,8,20

The earliest manifestations typically consist of erythematous, scaly, and sometimes atrophic macules or patches (Fig. 3A, B), which may vary considerably in size, shape, and distribution. Lesions most commonly occur in non-sun exposed areas, such as the buttocks, lower back, and thighs, although they may arise anywhere. Pruritus is uncommon, and lesions are usually asymptomatic. Over time, patches may evolve into infiltrated, well-demarcated plaques, often with a reddish-brown colour and variable scaling. The co-existence of patches and plaques in the same patient is a frequent finding.

Fig. 3.

Representative clinical presentations relevant to the diagnosis and differential diagnosis of early-stage mycosis fungoides. (A) Patch-stage MF with multiple faint erythematous patches involving the trunk and proximal extremities. (B) Classic MF plaque with well-demarcated erythematous, scaly lesions. (C) Folliculotropic MF presenting with grouped follicular papules and perifollicular infiltrated plaques. (D) Comedonal folliculotropic MF characterised by follicular plugging and comedone-like lesions on the trunk. (E) Urticarial plaques with central clearing and an annular configuration, predominantly affecting intertriginous areas. (F) Hypopigmented MF.

Folliculotropic MF typically presents with grouped follicular papules, acneiform lesions, or keratosis pilaris-like patches, often with associated alopecia, and most commonly involves the head and neck region (Fig. 3C, D).

Beyond the variants recognised by the World Health Organization (WHO) classification,¹ the literature describes several other clinical presentations that pose significant diagnostic challenges^1,20 (Fig. 3E, F). These less common forms may exhibit atypical morphology, distribution, or clinical course compared with conventional patch-plaque MF and often lack features that facilitate early recognition. Reported presentations include hypopigmented, hyperpigmented, poikilodermatous, purpuric, and ichthyosiform patterns, among others, frequently showing subtle, heterogeneous, or fluctuating features. Diagnosis is further complicated by substantial clinical overlap with a broad spectrum of benign inflammatory dermatoses, which may closely resemble MF in morphology and distribution but typically follow a more variable course and respond better to standard topical therapies. In contrast, MF lesions tend to be persistent and refractory to moderate-potency corticosteroids, often displaying subtle atrophy, poikiloderma, or a reddish-orange hue, with a predilection for non-sun-exposed sites. This overlap highlights the importance of careful longitudinal evaluation and close clinicopathological correlation to distinguish MF from its benign counterparts.^{1,4,6–8,20,28}

Molecular features and high-throughput analyses

Recent advances in molecular profiling have substantially improved our understanding of early-stage mycosis fungoides (MF), shedding light on its pathogenesis, molecular biomarkers, and potential therapeutic targets. A considerable body of work has focused on patch and plaque lesions, aiming not only to elucidate disease mechanisms but also to improve diagnostic accuracy and distinguish early-stage MF from other BIDs.

Transcriptomic analyses have demonstrated that early-stage MF exhibits distinct gene-expression patterns that clearly separate it from BIDs.^31–35 Alonso-Alonso et al.³⁵ identified differentially expressed genes between early-stage MF and BIDs, reporting overexpression of FGFR3, MAPK13, and LARG in BIDs, whereas IL15, LCP1, CD3E, CARD11, CD3D, TCRBC1/2, BIRC3, and CXCR4 showed significantly higher expression in neoplastic cells from early-stage MF.^3,35

In parallel, several investigations have examined microRNA (miRNA) expression profiles in early-stage MF compared with inflammatory dermatoses or advanced disease.^36–39 Sørensen et al.³⁷ demonstrated differential expression of miR-155, miR-142, and miR-146b between MF and psoriasis. Ralfkiaer et al.³⁶ further reported upregulation of miR-155, miR-146a/b-5p, miR-342-3p, and let-7i*, along with downregulation of miR-203 and miR-205, in MF relative to inflammatory dermatoses. Notably, several miRNAs were shared between MF and psoriasis, suggesting overlapping inflammatory pathways and early oncogenic signalling mechanisms. Complementing these findings, Manso et al.³⁸ identified 114 dysregulated miRNAs in early-stage MF compared with inflammatory dermatoses, including miR-181a and miR-146a, both previously associated with progression to advanced disease.

Progression of mycosis fungoides: from early to advanced stages

MF typically follows an indolent course in its early stages; however, a subset of patients undergoes a gradual but clinically significant progression to advanced plaque, tumour, erythrodermic, or extracutaneous disease. This transition reflects a complex interplay of histopathological evolution, increasing genomic instability, microenvironmental changes, and immunophenotypic aberrancies that accumulate over time. Understanding the biological mechanisms that drive this shift is crucial not only for identifying patients at risk and refining surveillance strategies, but also for distinguishing those with stable disease from those likely to develop more aggressive forms. Importantly, most disease-related mortality occurs after progression to tumour stage, erythroderma, or extracutaneous dissemination, underscoring the need for reliable predictors of progression even in early-stage disease.

Clinical and histopathological markers of progression

In early-stage MF, the risk of progression reflects the interplay between clinical burden, tissue architecture, cellular atypia, and immunophenotypic alterations. Clinically, extensive body surface area involvement, predominance of plaques over patches, folliculotropic MF, and head-and-neck localisation are consistently associated with less indolent behaviour. These features likely reflect increased neoplastic T-cell burden and deeper skin involvement, providing a biological link between clinical presentation and underlying tumour expansion.^1,4

Histopathology is central to risk evaluation, though its significance is best interpreted alongside clinical findings. Dense or band-like dermal lymphoid infiltrates, folliculotropism, Pautrier microabscesses, adnexotropic involvement, and prominent cerebriform atypia indicate a more advanced biological phenotype. These features, typically associated with progression from patch to plaque stages, are often accompanied by partial loss of epidermotropism. Such patterns may reflect early architectural remodelling and microenvironmental adaptation, representing precursor steps towards tumour-stage MF.^1,5

Immunophenotypic alterations further refine prognostication. High GATA3 expression, >30% CD7 loss, elevated Ki-67 (>20%), and CD30 positivity (>10%) have been associated with disease progression, particularly the development of tumoral lesions and nodal dissemination. Upregulation of CXCL13 similarly identifies lesions with increased trafficking potential and microenvironmental remodelling. Conversely, loss of CD2 or CD5 does not appear to predict adverse outcomes, underscoring the heterogeneity of T-cell marker deregulation in MF. Although the prognostic role of tumour-infiltrating B cells remains uncertain, these clinical, histological, and immunophenotypic features may aid in risk stratification of early-stage MF.^1,40

Additionally, large-cell transformation (LCT) represents a histologically and clinically distinct manifestation of disease progression. Defined by the emergence of large, atypical T cells, LCT often presents as rapidly enlarging plaques or tumours and may involve multiple cutaneous or extracutaneous sites. Although more common in advanced MF, up to one-third of cases arise in early-stage disease, and its presence is associated with increased progression risk and poorer prognosis. Median overall survival following transformation is approximately 3.5 years, with unifocal cutaneous lesions showing more favourable outcomes than multifocal or extracutaneous involvement, highlighting the importance of recognising LCT as a marker of disease advancement.^1,24

Molecular predictors of progression

Several molecular studies have focused on early-stage MF, particularly in patch and plaque lesions, aiming to elucidate disease mechanisms and identify predictors of progression. Transcriptomic analyses have shown that early-stage MF displays gene expression patterns distinct from those of advanced-stage disease.^31–34 Early lesions typically contain a low burden of malignant T cells and are characterised by a microenvironment dominated by cytotoxic CD8⁺ resident memory T cells (TRM), Th1-polarised CD4⁺ T cells, NK cells, and activated dendritic cells. Keratinocytes and dermal fibroblasts actively support this protective anti-tumour environment through the production of Th1-attracting chemokines, including CXCL9, CXCL10, and CXCL11, which promote recruitment and retention of cytotoxic and Th1-type lymphocytes within the skin. In this early phase, immune surveillance appears to partially restrain malignant T-cell expansion.^3,32,40

As the disease progresses, this balance gradually shifts. Activated keratinocytes begin producing Th2-attracting chemokines, including CCL17, CCL22, and CCL27, favouring recruitment of Th2 cells and weakening the Th1-mediated anti-tumour response. Cancer-associated fibroblasts (CAFs) further reinforce this process through periostin secretion, which induces thymic stromal lymphopoietin (TSLP) production by keratinocytes, establishing a feedback loop that sustains Th2 polarisation. Dendritic cells amplify this environment by promoting antigen-driven activation and Th2-type inflammation. Progression is therefore characterised by a reduction in cytotoxic CD8⁺ T cells, expansion of atypical CD4⁺ T cells and M2 macrophages, and a shift towards a tumour-promoting, Th2-dominant immune landscape.^3,31,34 Additional stromal and immune components, including dermal dendritic cells, activated keratinocytes and epidermal Langerhans cells, have also been shown to support malignant T-cell expansion and contribute to disease aggressiveness. Changes in the frequency and distribution of Tregs and Th17 cells have likewise been associated with disease progression.³

This immunological reprogramming is reflected in cytokine profiles. Early-stage MF shows high expression of Th1 cytokines, such as IL-2 and IFN-γ, whereas advanced disease is characterised by increased expression of IL-5, IL-10, IL-13, and IL-17. Notably, IL-4 and IL-13 act synergistically to enhance tumour cell proliferation, underscoring the functional relevance of this shift.^3,34

Whole-transcriptome and spatial transcriptomics approaches further support these observations. Early-stage MF exhibits gene expression patterns distinct from those of advanced disease. During progression from patch to tumour-stage, there is a progressive activation of MAPK/ERK signalling, CD28 co-stimulation, B-cell-related pathways, PI3K–AKT–mTOR signalling, and Th2/Th9-associated STAT3/5/6 pathways.^3,32–34,40 Spatial transcriptomics indicates that stromal remodelling begins early, with increased CAFs and macrophage polarisation changes already detectable in plaque lesions (Fig. 4).^3,31

Fig. 4.

Molecular mechanisms driving the progression of early-stage MF. The figure illustrates the key molecular alterations underlying the transition of MF from its early phases to advanced disease, including activation of oncogenic signalling pathways, overexpression of neoplastic markers, dysregulation of microRNAs, genomic instability and somatic mutations, and remodelling of the lesional microenvironment. These changes collectively promote immune escape and the uncontrolled expansion of malignant T cells.

Rebollo-González et al.⁴⁰ identified a transcriptomic signature predictive of progression, highlighting overexpression of CD30, CXCL13, and BIRC3 as biomarkers associated with disease advancement. These markers align with key biological pathways driving MF pathogenesis, collectively indicating early activation, enhanced survival capacity, and microenvironmental remodelling in lesions that subsequently progress. Their combined expression suggests increased proliferative activity of the neoplastic T-cell clone, recruitment of B cells and TFH-associated signalling that may support ectopic lymphoid structures, and activation of NF-κB-dependent anti-apoptotic mechanisms that favour clonal persistence and expansion.^5,32,40,41 Additional markers linked to future progression include KIR3DL2, MCM7, PCNA, BIRC5, IL-4, and IL-13, together with reduced STAT4 and FOXP3 expression (Fig. 4). These findings reflect an early shift in immune regulation within lesional skin, characterised by Th2-dominant signalling, impaired Th1 responses, and disruption of regulatory networks. Collectively, these alterations indicate that immune dysregulation, pro-survival pathway activation, and aberrant cytokine signalling arise early in the disease course, contributing to the increased likelihood of progression in certain lesions. Although these signatures show promising predictive value, validation in large longitudinal cohorts remains necessary before clinical implementation.^3,32,40

At the signalling level, PI3K and RAS pathways appear central to progression. Several deregulated genes, including members of the FGF family, EFNAs, HGF, PGF, SOS1, and KIT, suggest functional cross-talk between these cascades. Cooperative activation of RAS and PI3K may be required for transition to tumour-stage MF, potentially explaining why such progression remains relatively infrequent.^33,34

miRNA profiling has revealed that early-stage MF, even at the patch stage, exhibits a distinct miRNA signature compared with advanced disease.^36–39 Sorensen et al.³⁷ demonstrated differential expression of miR-21 and miR-22 between patch and plaque lesions, implicating these miRNAs in disease progression. Manso et al.³⁸ identified dysregulated miRNAs between early and advanced stages, including miR-181a and miR-146a, which were associated with progression. Avgeros et al.⁴² reported that circulating levels of miR-146a and miR-155 were elevated even in early-stage MF compared with healthy controls and increased with disease severity, correlating with specific polymorphisms (rs2910164 and rs767649). These findings suggest the potential utility of miRNAs as non-invasive biomarkers for diagnosis, prognosis, and disease monitoring (Fig. 4).

Genomic analyses have identified recurrent alterations associated with early-stage MF. Single nucleotide polymorphisms (SNPs) in genes such as GZMB, HLA-DRB1, CD103, NOTCH1, LAG3, NR4A2, and CD26L, many of which regulate skin-resident memory T-cell homeostasis, may increase susceptibility to disease initiation.^43–46 Fléchon et al.⁴⁴ reported that deletion of 17q11.2 (affecting SUZ12 and NF1) was associated with lower-risk stages (IA–IIA), whereas deletion of 6q16.3 (TNFAIP3) correlated with more rapid progression. Mutations in JUNB and TET2 were linked to high-risk, advanced-stage disease (Fig. 4).

Wehkamp et al.⁴⁷ applied EuroClonality-NGS assays to detect circulating tumour clones via T-cell receptor gene rearrangements in patients with MF, demonstrating that most MF patients harbour identifiable clonal signatures in peripheral blood that can distinguish malignant from inflammatory conditions. Although some clonal TCR rearrangements may also be seen in benign inflammatory dermatoses, the presence of tumour-specific TCR clones in blood represents a promising tool for both supporting diagnosis and monitoring disease burden and potential progression in early-stage MF. Supporting this, De Masson et al.⁴⁸ showed that in cutaneous lesions of early-stage MF, a high tumour clone frequency (TCF > 25%) detected by high-throughput TCR-β sequencing was strongly associated with reduced progression-free and overall survival, identifying patients at higher risk of disease progression even in initial stages. This highlights the potential of NGS-based TCR clonality assessment as both a diagnostic and prognostic biomarker in early-stage MF.

Together, these data indicate that MF progression reflects coordinated immune reprogramming, microenvironmental remodelling, genomic instability, and activation of oncogenic signalling pathways. While multiple molecular signatures show predictive potential, prospective validation in well-defined cohorts is essential before integration into routine clinical practice.

Prognostic tools and risk stratification

In early-stage MF, progression risk is determined by a combination of clinical, histopathological, and immunophenotypic features, and can be further stratified using modern prognostic tools.

Several prognostic indices have been developed to better stratify patients with early-stage MF, aiming to integrate clinical, laboratory, and pathological data into actionable tools for predicting disease progression. These indices allow clinicians to identify patients at higher risk of progression, guiding surveillance and early intervention strategies. The modified Severity-Weighted Assessment Tool (mSWAT), although initially designed for response assessment, also provides prognostic value: higher baseline mSWAT scores and failure to achieve significant reduction after therapy predict faster progression. The Cutaneous Lymphoma International Prognostic Index (CLIPi) is one of the most validated prognostic models for early-stage MF, stratifying patients into low-, intermediate-, and high-risk groups based on male sex, age ≥60 years, stage IB–IIA, and presence of plaque lesions; it identifies patients at increased risk of progression and disease-related mortality. The PROCLIPI (Prospective International Cohort Study) project represents the largest prospective effort to define predictive models specifically for early-stage MF. Preliminary analyses highlight that folliculotropic variants, plaques, >10% BSA involvement, and, in some cohorts, elevated LDH are strong predictors of early progression. Complementing CLIPi and mSWAT, PROCLIPI integrates clinical, laboratory, and molecular variables to refine risk categories, enabling objective stratification of patients and supporting early therapeutic intervention in those at higher risk.^49,50

Conclusions – future research directions

Early-stage MF remains one of the most challenging entities in cutaneous pathology, primarily because its clinical and histological subtleties so often overlap with BIDs. A comprehensive diagnosis of early-stage MF integrates clinical assessment with histopathological, immunophenotypic, molecular, and immunohistochemical findings. Maintaining a high clinical index of suspicion for this “great imitator” is crucial to ensure timely, accurate diagnosis and appropriate patient management.

Looking ahead, the field is entering a transformative phase in which molecular and spatially resolved technologies are beginning to illuminate the earliest biological events that drive MF initiation and progression. Deciphering the interplay between neoplastic T cells, immune regulators, and stromal elements will be central to defining biomarkers that can distinguish MF from inflammatory mimickers and identify lesions destined for progression. High-throughput transcriptomics, multiplex immunohistochemistry, spatial transcriptomics, and single-cell analyses hold particular promise for uncovering microenvironmental signatures, clonal dynamics, and signalling pathways that may serve both as diagnostic adjuncts and therapeutic targets.

Future research should prioritise large, longitudinal, multicentre cohorts that integrate clinical phenotyping with deep molecular profiling to generate validated prognostic models capable of guiding clinical decision-making. Equally important will be the development of standardised molecular criteria that can complement traditional pathology, reducing diagnostic ambiguity and allowing earlier deployment of personalised surveillance and treatment strategies.

By merging classical anatomical pathology with cutting-edge molecular tools, the next decade has the potential to redefine early-stage MF, not merely as a challenging diagnosis, but as an increasingly predictable, measurable, and precision medicine-amenable disease.

Author contributions

S.M.R.-P. conceived the study, provided overall supervision, and oversaw the methodological and conceptual framework of the review. M.R.-G. drafted the manuscript, prepared the original draft, and designed all figures included in the manuscript. R.M., M.R. and J.F.D.-A. conducted the critical appraisal and synthesis of the molecular data. J.T.-C. contributed to the clinical section and its interpretation. F.J.D.d.l.P. was responsible for the histopathological section and provided expert review of the corresponding content. The immunohistochemistry case figures were designed by S.M.R.-P. and F.J.D.d.l.P. All authors critically reviewed the manuscript, contributed to its refinement, and approved the final version.

Informed consent

Not applicable.

Ethical considerations

Not applicable

Funding

This study was supported by grants from the Instituto de Salud Carlos III (ISCIII), Spanish Ministry of Economy and Competitiveness. It was co-funded by the European Union (FEDER) (ERDF/ESF, “A way to make Europe”/“Investing in your future” (MINECO, ISCIII), including support from the European Union Next Generation EU initiative through the Recovery and Resilience Mechanism (RRM) (Plan Nacional I+D+I: PI21/01724, 15826/004, 41163/005 and PMP21/00015). Additional funding was provided by the AECC, the Madrid Autonomous Region, and STARTUP2020/L2566. M.R.-G is the recipient of a predoctoral scholarship “Ayudas para la contratación de personal investigador predoctoral en formación 2024” (PIPF-2024/SAL-GL-35294). J.F.D.-A. is the recipient of a Fundación Conchita Rábago (FCR, IIS-FJD) predoctoral scholarship.

Conflicts of interest

The authors declare no conflicts of interest.