Peripheral T-cell lymphoma, not otherwise specified (PTCL-NOS), is the most common histological subtype of PTCL (1); its clinical, immunophenotypic and genetic features are highly heterogeneous. The global incidence of PTCL-NOS is 0.5–1.7 cases per 100,000 person-years (2,3). The disease shows marked geographic variation; in China, PTCL-NOS accounts for 23–27% of all non-Hodgkin lymphoma cases, a figure notably higher than the 10% reported in Western countries (4). In the absence of distinctive immunophenotypic and molecular signatures, diagnosis is based on exclusion of other diseases (5). The clinical course is characteristically aggressive, with most patients presenting with advanced stage (III/IV) disease (5). Extranodal involvement is seen in ~50% of cases, most frequently in the bone marrow, skin and gastrointestinal tract (6). Typical laboratory abnormalities include anaemia, thrombocytopenia, elevated lactate dehydrogenase, increased β₂-microglobulin, eosinophilia and hypergammaglobulinemia (7). Histologically, diffuse effacement of lymph-node architecture is the most common pattern; residual B cells are often seen at the periphery of the infiltrate, and some cases show selective interfollicular or paracortical involvement. Tumour cells are not morphologically distinctive, consisting of medium- to large-sized pleomorphic lymphocytes with irregular nuclei, admixed with reactive elements, including eosinophils, neutrophils, histiocytes and plasma cells. Immunohistochemistry (IHC) may reveal loss of pan-T-cell antigens, most commonly CD5 and CD7. T-follicular helper (TFH)-associated markers [BCL6, CD10, inducible T-cell costimulator/CD278, serum amyloid P component, C-X-C motif chemokine ligand 13 and C-C chemokine receptor type 5 (CCR5)] can aid in identifying TFH-derived lymphoma; however, isolated positivity for a single TFH marker does not exclude a diagnosis of PTCL-NOS (8). PTCL-NOS is associated with poor outcomes, and no standard-of-care therapy exists. Current management still borrows from B-cell lymphoma protocols, with the cyclophosphamide, vincristine, doxorubicin and prednisone (CHOP) regimen remaining the default first-line regimen. Although the initial objective response rate is modest, early relapse is common. For patients treated with CHOP regimen, the objective response rate (ORR) is 56%, and nearly one-half of the patients experienced relapse or progression within 6 months after primary therapy, with a 5-year progression-free survival (PFS) rate of only 29% (6,9,10). In recent years, increasing attention in lymphoma research has shifted toward molecular prognostic biomarkers and pathogenic mechanisms (11,12). In parallel, a wave of novel targeted agents has emerged for PTCL, including chidamide, romidepsin and brentuximab vedotin (BV), all of which have conferred measurable clinical benefit (8).

Nevertheless, the long-term outlook for relapsed/refractory PTCL-NOS remains poor. As the disease is rare, its molecular pathogenesis and mechanisms of drug resistance are still inadequately defined, and research progress has been slow; consequently, these patients face a therapeutic impasse. The present study was therefore designed to delineate the clinical characteristics of PTCL-NOS, identify potential prognostic determinants, and provide data to facilitate an earlier diagnosis and improve long-term survival.

PTCL-NOS is the most common subtype of PTCL, accounting for 21–27% of all PTCL cases (22). Although PTCL-NOS remains defined as a heterogeneous collection of tumours that cannot be assigned to any other recognized PTCL category, recent advances in PTCL genomics have led to the reclassification of cases exhibiting a TFH phenotype or Epstein-Barr virus (EBV) positivity as distinct entities, thereby narrowing the diagnostic scope of PTCL-NOS (23).

In the present study, the clinical characteristics of 30 patients with PTCL-NOS were analysed; the median age at diagnosis was 59 years, and the disease was more common among men (73.3%). The majority of patients presented with painless peripheral lymphadenopathy, and approximately one-half (43%) had B symptoms (fever, night sweats or weight loss). Occasional co-occurrence of hemophagocytic syndrome and myelofibrosis was also observed, which is consistent with a previous report (6). PTCL-NOS typically arises in the lymph nodes, but the present cohort also included rare primary extranodal cases involving the gastrointestinal tract, skin and muscle. Previous research has shown that the primary site influences outcome: Tumours originating in the kidney or gastrointestinal tract are associated with poorer survival than nodal disease, whereas primary bone, skin or soft-tissue involvement confers a better prognosis (24).

Immunophenotypically, all patients in the present cohort exhibited partial or complete loss/downregulation of pan-T-cell antigens, most frequently CD7 and CD5. Overall, 60% of cases were positive for the cytotoxic marker TIA-1, a figure considerably higher than the 32% reported in Western countries (North America and Europe) (6) and further evidence that cytotoxic PTCL-NOS (cPTCL-NOS) is more common in Asian populations (25). cPTCL-NOS is defined by expression of ≥1 cytotoxic marker (TIA-1, granzyme B or perforin) in ≥50% of tumour cells (25). These patients often have an immunocompromised background, such as a history of malignancy or autoimmune disease, and the pathogenesis is linked to mutations in epigenetic regulators, including tet methylcytosine dioxygenase 2 and DNA methyltransferase 3α. Studies have shown that cPTCL-NOS typically expresses the T-helper 1-associated transcription factors TBX21 and C-X-C chemokine receptor type 3 (CXCR3), and belongs to the PTCL-TBX21 subgroup, which is associated with a poor prognosis (26). PTCL-TBX21 is driven by the transcription factor TBX21 and characterized by aberrant activation of the STAT3 pathway; another subgroup identified by gene expression profiling is PTCL-GATA3, which is driven by the transcription factor GATA3. Compared with PTCL-TBX21, PTCL-GATA3 exhibits a more aberrant genomic expression profile and is associated with PI3K pathway activation, with a worse overall prognosis (27). An IHC algorithm using TBX21/CXCR3 and GATA3/CCR4 as markers has been developed to distinguish PTCL-TBX21 and PTCL-GATA3 in clinical practice (28). Due to the lack of fresh, formalin-fixed, paraffin-embedded pathological samples, it was not possible to perform supplementary IHC staining for the aforementioned markers and gene-expression profiling in the present study, thereby limiting the ability to validate the prognostic value of molecular subtypes and specific gene mutations in cPTCL-NOS in the real-world setting. Notably, previous cohort studies of cPTCL-NOS primarily included EBV-positive cases (28,29); these have now been reclassified as distinct entities. Large-scale investigations are urgently needed to determine whether EBV-negative cPTCL-NOS exhibits clinical features and survival outcomes that differ from those of non-cPTCL-NOS (30). In the present study, the CD30 positivity rate was 53.8%, which is higher than that reported in an earlier study (22), likely reflecting differences in the CD30 positivity cutoff and the limited sample size. This finding indicates that BV, an antibody-drug conjugate targeting CD30, could be a therapeutic option for these patients.

Currently, the clinical prognostic indices most frequently used for PTCL-NOS are the IPI, the PIT and the modified PIT; all three models stratify patients according to age, ECOG PS, LDH level, number of extranodal sites and Ki-67 expression (6,19,31,32). The present study confirmed that an ECOG PS >1, bone marrow involvement and platelet count <150×109/l are independent predictors of OS, which is consistent with previous reports (6,31,32). In addition to bone marrow involvement, albumin levels <35 g/l were an independent predictor of PFS. Low serum albumin levels have been linked to poor outcomes in a variety of malignancies, including gastric, small-cell lung and breast cancer (33–35), and its prognostic value across PTCL subtypes has been confirmed by multiple studies (36–39). Albumin reflects not only systemic nutritional status but also the inflammatory burden and tumour load. Pro-inflammatory cytokines released within the tumour microenvironment, such as IL-6 and tumour necrosis factor-α, suppress hepatic albumin synthesis, whereas increased capillary leakage accelerates the development of hypoalbuminemia. Raj et al (40) analysed 149 patients with large B-cell lymphoma and demonstrated that C-reactive protein and IL-6 levels were inversely associated with serum albumin level, confirming that albumin behaves as a negative acute-phase protein that mirrors the high inflammatory state of lymphoma. Further mechanistic studies have shown that IL-6 is a key component of a positive inflammatory feedback loop, activating the JAK1-STAT3 pathway, and thereby promoting proliferation and migration of PTCL cells (41,42). Federico et al (36) incorporated the prognostic value of serum albumin into a T-cell scoring system that combines serum albumin level, ECOG PS, disease stage and absolute neutrophil count, and validated its ability to predict both OS and PFS. Although this model, derived from a large prospective dataset, achieved greater predictive accuracy than earlier indices, it remains limited to clinical variables and offers limited guidance for treatment selection. In the present study, albumin was measured only once prior to treatment initiation, precluding any dynamic assessment of its relationship with therapeutic response. Owing to the retrospective nature of the data, inflammatory markers such as C-reactive protein and IL-6 were excluded from the model; therefore, the proposed link between hypoalbuminemia and systemic inflammation could not be formally validated. Large prospective cohorts are still needed to clarify how pretreatment albumin levels are associated with nutritional status and inflammatory burden in real-world PTCL-NOS, with the ultimate goal of developing robust predictive algorithms that can reliably identify high-risk patients.

In the present cohort, anthracycline-based chemotherapy was the primary strategy, yielding an ORR of 56.7% and a 5-year OS of 44.8%, figures that mirror historical data (6). CHOP-like regimens perform poorly in PTCL-NOS. Although first-line response rates are acceptable, early relapse often occurs, and long-term outlook remains poor. The value of anthracyclines is itself debated; several studies have found no notable survival difference between anthracycline-containing and anthracycline-free schedules (6,43). In the present study, the outcomes between patients receiving CHOP and those receiving CHOPE were compared, and no statistically significant difference in either OS or PFS was observed. A previous report has suggested that CHOPE may improve PFS rate in patients <60 years old, possibly as elderly or frail individuals are less tolerant of the myelosuppressive effects of etoposide (44); therefore, age-stratified analyses are required to clarify the true benefit of the CHOPE regimen.

Chidamide, a selective histone deacetylase inhibitor developed in China, is now widely used to manage PTCL and shows favourable activity (45). In the present study, patients who received chidamide exhibited higher 3-year OS and PFS rates compared with those who did not receive the treatment, with a notable trend toward improved PFS. However, this difference did not reach statistical significance, likely due to the limited sample size. The small sample size also precluded stratified analysis of outcomes based on different chidamide treatment strategies, such as up-front combination therapy vs. maintenance therapy. Additionally, clinicians tend to initiate chidamide treatment earlier in patients deemed high-risk. Furthermore, financial constraints prevent some patients from completing the planned number of treatment cycles. These factors may bias the efficacy estimates toward the null, potentially obscuring the true therapeutic benefits of chidamide. The small cohort and variable follow-up duration further reduced the statistical power in the present study. Consequently, the negative result observed did not refute the therapeutic value of chidamide, rather, it simply documents the heterogeneity of its real-world use. Large prospective studies are required to define the precise role of chidamide as first-line and maintenance therapy, and to determine whether molecular subgroups derive differential benefit. BV, an anti-CD30 antibody-drug conjugate, has demonstrated therapeutic potential in CD30-positive lymphoma (46,47). In the present study, 1 patient who relapsed (repeat lymph-node biopsy confirming CD30 positivity) received BV combined with chemotherapy as salvage therapy, and achieved PR, before proceeding to allo-HSCT. This single-case experience suggests that BV is active in PTCL-NOS. Beyond improving drug accessibility (such as reducing economic burden), an evidence-based CD30 cutoff must be established, and prospective real-world studies on the efficacy and safety of BV in China are urgently needed to inform clinical decision-making.

A total of 3 patients in the present study underwent autologous stem-cell transplantation after achieving CR, and all remain alive at the latest follow-up. The precise value of upfront autologous stem-cell transplantation in PTCL-NOS is still contested. A large retrospective study showed improved OS when patients aged <65 years with stage II–IV disease received autologous stem-cell transplantation as consolidation (48), whereas a prospective trial found no survival benefit for sequential autologous stem-cell transplantation in first-remission PTCL-NOS (49). Although existing data are conflicting, and large-scale studies specific to the PTCL-NOS subtype are lacking, current guidelines and clinical practice continue to favour sequential autologous stem-cell transplantation for chemotherapy-sensitive patients, on the assumption that it may achieve long-term disease control.

Patients with relapsed or refractory PTCL-NOS have poor outcomes, effective agents are limited, and they remain in a challenging therapeutic situation. In the present cohort, the 3-year OS was only 21.8%, and 4 of the 21 patients died from rapid progression, within a median OS time of 0.50 months, without ever receiving salvage therapy. Although the sample is too small to identify statistically robust high-risk features, all four had high stage disease, serous effusion, anaemia and a Ki-67 index ≥70%, suggesting that an inherently aggressive biology, rather than poor PS, drives the fulminant course. Larger studies integrated with molecular genetics are required to determine whether an ultra-high-risk molecular subset exists. Salvage regimens in the present study were highly heterogeneous; conventional chemotherapy (GDP and ICE) conferred only a modest survival benefit, which is consistent with previous reports (50–52). Notably, the mere opportunity to receive salvage therapy emerged as a survival ‘watershed’, implying that published response rates may be systemically overestimated. Additionally, the use of targeted agents (chidamide and BV) in Chinese patients with relapsed/refractory PTCL-NOS was piloted, providing preliminary experience for individualized, targeted combination strategies. Transplant-related mortality is undeniable, yet allo-HSCT remains the only curative option for relapsed/refractory PTCL-NOS. In the present study, only 2 patients were bridged to allo-HSCT; most younger individuals never achieved a response adequate for transplant, and the need to locate a suitable donor and secure costly procedures before further progression created additional, real-world barriers. More effective pre-transplant salvage regimens are urgently required, while frail patients who are ineligible for transplants need lower-intensity alternatives that can meaningfully prolong survival. As the genetic understanding of PTCL pathogenesis improves, novel targeted agents, such as the JAK1 inhibitor golidocitinib, the PI3K inhibitor linperlisib and the programmed cell death protein 1 inhibitor pembrolizumab, are emerging, offering patients with relapsed/refractory PTCL-NOS additional therapeutic options. Thus, patients with PTCL-NOS may achieve sustained remission or even disease-free status with effective targeted therapies.

In the present study, treatment-related toxicities were generally manageable; the most common grade 3–4 adverse events were myelosuppression-neutropenia (30%), anaemia (16.7%) and thrombocytopenia (20%). These rates are largely consistent with those of a previous report (53). Pneumonia was the leading cause of treatment delay, and the two grade 3–4 pulmonary infections were both neutropenia-related. Clearly, maintaining chemotherapy intensity to achieve remission while avoiding severe myelosuppression-related complications remains a formidable clinical challenge. Prophylactic G-CSF can reduce the incidence of serious infections and maintain cycle schedules, but cumulative toxicity, dynamic blood count trends and individual infection histories must all be considered. Therefore, a multidisciplinary consultation should be sought promptly when necessary. Overall, the regimens used in the present study were considered safe, and no treatment-related deaths occurred.

The present study has several important limitations that restrict the generalizability of the findings. First, the overall sample size and the sizes of individual subgroups are small, limiting statistical power and increasing the risk of type II errors; comparisons of survival between groups must therefore be interpreted with caution. Second, the study is a single-centre retrospective analysis and is subject to inherent selection and recall biases; although baseline clinical characteristics were compared across all subgroups, potential confounders such as socioeconomic factors, patient treatment preferences and individual physician choices may still have influenced survival outcomes. Third, the absence of an external validation cohort means that the results are exploratory only; they provide clinical evidence and generate preliminary hypotheses for this rare disease, but a large, multicentre, prospective study is required for confirmation.

In summary, PTCL-NOS predominantly affects middle-aged and elderly men, follows an aggressive clinical course and is associated with poor overall outcomes. Conventional CHOP-like regimens provide sub-optimal disease control; relapse is common and robust prognostic tools are lacking. Multicentre prospective studies with larger patient cohorts are urgently required, together with comprehensive mapping of the PTCL-NOS molecular landscape, to elucidate the biological basis of its clinical heterogeneity and to develop effective targeted combination strategies that improve long-term survival for patients with this disorder.