The data from patients with nTFHL-F or nTFHL-NOS were obtained from six large comprehensive tertiary hospitals in China spanning from December 2017 to October 2023 ( Supplementary Table S1 for participating institutions). The clinical data are summarized in Table 1 . Patients with nTFHL-AI diagnosed at the First Affiliated Hospital of Zhengzhou University during the same period were also included for comparison. The follow-up cutoff date was October 30, 2023. The study was conducted in accordance with the Declaration of Helsinki and approved by the ethics review board.

Expert hematopathologists from the pathology department of each participating center reviewed the pathological specimens, including slides stained with hematoxylin-eosin (H&E), immunohistochemistry, and in situ hybridization for Epstein–Barr virus encoded RNA (EBER-ISH), at the time of enrollment to confirm the diagnosis. The diagnosis of nTFHL-AI was defined as the destruction of lymph nodal structure accompanied by dendritic hyperplasia of high endothelial venules (HEVs), the proliferation and expansion of follicular dendritic cell (FDC) networks, and a polymorphic inflammatory background containing histiocytes, plasma cells, and eosinophils ( Figure 1 ). The diagnosis of nTFHL-F required a follicular growth pattern and FDC meshwork associated with follicles without extrafollicular FDC expansion. The follicles were filled with abnormal T cells expressing TFH markers ( Figure 2 ). The diagnosis of nTFHL-NOS was assigned to CD4+ nPTCL with a TFH phenotype that did not meet the criteria for nTFHL-AI and nTFHL-F. This condition usually refers to the absence of dendritic hyperplasia of HEVs, proliferative expansion of the FDC network, and a complex inflammatory background ( Figure 3 ). Comprehensive diagnostic criteria were applied, incorporating clinical presentation, morphological features, and immunohistochemical findings. Genetic rearrangement testing was performed when diagnostic uncertainty persisted to facilitate a definitive diagnosis (2, 10–12).

The specimen was fixed in 3.7% neutral formaldehyde, routinely deparaffinized, embedded in paraffin, and cut into 3 μm sections for histological examination using hematoxylin and eosin staining.Immunohistochemical staining was performed using the following antibodies: CD3, CD20, CD5, CD4, CD8, CD10, CD21, BCL-6, CXCL13, PD1, CD30, and Ki-67. The basic information of the antibodies is summarized in the Supplementary Table S2 . Positive results were determined as follows: PD-1 exhibited cytoplasmic brown-yellow staining, CD10 showed cytoplasmic/membranous brown-yellow staining, bcl-6 exhibited nuclear brown-yellow staining, and CXCL-13 exhibited perinuclear punctate brown-yellow granular staining. The positivity for TFH markers was defined as the expression in at least 10% of tumor cells.

EBV status was determined by ISH to detect EBV-encoded RNA 1 and 2 (EBER1/2s) using peroxidase-labeled probes ISH-7001UM (Beijing Zhongshan Golden Bridge Biotechnology, Beijing, China). Tissues from an EBV-positive NK/T cell lymphoma were used as a positive control, and PBS buffer as a negative control in place of primary antibody or hybridization solution. Positive cells (PCs) were defined as the presence of moderate or higher intensity staining (brown/dark brown) in the nucleus. The number of PCs was recorded in high-power fields (HPFs). we defined EBER-positive neoplastic cells per high-power field (HPF) exceeded 50 (> 50/HPF) as EBER(+). Samples with scattered (the number of positive neoplastic cells in the highest density region was less than 50/HPF), inadequate, or equivocal positive neoplastic cells (a cutoff value < 50/HPF) were as EBER (–).

Polymerase chain reaction (PCR) amplification of T-cell receptor genes and immunoglobulin gene rearrangements were performed for detecting monoclonality when necessary following the BIOMED-2 protocol as described previously (13). DNA was extracted from the QIAamp DNA Mini Kit (QIAGEN, Hamburg, Germany) following the manufacturer’s protocol. Commercial BIOMED-2 multiplex PCR kits (Righton Gene, Shanghai, China) were used for PCR amplification of TCR and IG gene rearrangements. After separation by capillary electrophoresis, PCR products were subjected to GeneScan analysis on an ABI 9700Genetic Analyzer (Applied Biosystems, CA, USA) and analyzed using the GeneMapper software (Version 4.0; Applied Biosystems, CA, USA).

Clinical data were collected through the use of case report forms. The clinical data included age, sex, B symptoms, Eastern Cooperative Oncology Group (ECOG) performance status, immune inflammatory-related symptoms, Ann Arbor stage, extranodal involvement sites, the International Prognostic Index (IPI), the prognostic index for peripheral T-cell lymphoma unspecified (PIT), treatment methods, treatment response, survival status, and cause of death. The recorded laboratory data included baseline complete blood cell count, albumin level, globulin level, serum lactate dehydrogenase (LDH) level, C-reactive protein (CRP) level, serum immunoglobulin level, Coombs test, and ANA test. Bone marrow involvement was diagnosed through bone marrow biopsy. The involvement of the spleen and other extranodal sites was determined using diagnostic tools such as CT, enhanced CT, and PET-CT. Treatment response was evaluated according to the 2014 Lugano classification criteria (14). The response was categorized as complete remission (CR), partial remission (PR), stable disease (SD), or progressive disease (PD). The objective response rate (ORR) was defined as the proportion of patients who achieved PR and CR.

Quantitative data are expressed as either the mean ± standard deviation (SD) or median with interquartile range (IQR). Intergroup comparisons were conducted using Student’s t test, the Mann−Whitney U test, or the Kruskal−Wallis test, as appropriate. Categorical data are reported as percentages or proportions, and comparisons were made using either Pearson’s chi-square test or Fisher’s exact test. Overall survival (OS) was defined as the time interval from diagnosis to either death from any cause or the last follow-up. The time to disease progression (TTP) was defined as the duration from the initiation of treatment to disease progression. Progression-free survival (PFS) was defined as the period from the commencement of treatment to either disease progression or death from any cause. Survival analysis was conducted utilizing the Kaplan−Meier method, and the log-rank test was used to compare survival rates between the two groups. A P value < 0.05 was considered to indicate statistical significance. The statistical analysis was conducted using SPSS software, specifically version 26.0.

Table 2 summarizes the baseline characteristics of the entire study population. The nTFHL-NOS patients were younger at disease onset (62.5 vs. 59.0 years, P=0.134) than the nTFHL-AI patients were, and they presented less frequently with ECOG score > 1, B symptoms, rash, and serosal effusion. Laboratory and imaging tests revealed a decreased incidence of anemia, decreased albumin levels, elevated globulin levels, elevated LDH levels, elevated CRP levels, increased polyclonal serum immunoglobulin levels, and advanced disease stage. Most patients demonstrated low to intermediate risk according to both the IPI and the PIT score, and these differences were statistically significant. However, there were no statistically significant differences in clinical presentation, laboratory or imaging findings, staging, IPI score, or PIT score between the nTFHL-F cohort and the nTFHL-AI cohort as well as the nTFHL-NOS cohort.

A majority of the patients underwent CHOP-like induction therapy. The ORR for nTFHL-AI, nTFHL-F, and nTFHL-NOS were 41% (25/64), 29% (2/7), and 75% (9/12), respectively ( Table 3 ). The GDPT and C-PET regimens demonstrated improvements in the ORR (P=0.0141), OS (P=0.0417, Figure 4G ), and PFS (P=0.0258, Figure 4H ) for nTFHLs compared to the CHOP-like regimen. The median OS for the CHOP-like, GDPT, and C-PET regimens were 12 months (95% CI 6.9–17.1), 20.0 months (95% CI 6.9–33.1), and 24.6 months (95% CI undermined), respectively, while the median PFS were 5 months (95% CI 3.4–6.6), 9.6 months (95% CI 0.0–22.0), and 10 months (95% CI 3.0–17.0).

With a median follow-up of 36.1 months, the median OS for patients with nTFHL-AI, nTFHL-F, and nTFHL-NOS were 10 months (95% CI 4.5-15.5), 5 months (95% CI 0.7-9.3), and; 14.2 months (95% CI 12.1-16.3), respectively ( Figure 4A ), and the median TTP was 5 months (95% CI 3.8-9.2), 3 months (95% CI 0.0-6.7), and 14 months (95% CI 3.8-24.2). The results of the log-rank test for the nTFHL-AI, nTFHL-F, and nTFHL-NOS subtypes of TTP survival curves showed an overall difference in prognosis (P=0.0173, Figure 4B ). The survival of the nTFHL-NOS subtype was significantly better than that of the nTFHL-AI subtype (P=0.0155, Figure 4B ) and the nTFHL-F subtype (P=0.0075, Figure 4B ). Additionally, in the group of patients receiving induction therapy with CHOP-like regimens, significant differences were observed in OS (P=0.0134, Figure 4C ) and TTP (P=0.0205, Figure 4D ) among the three subtypes. Both the IPI and PIT score retained prognostic value in assessing the outcome of nTFHLs (IPI, P<0.001; Figure 4E ; PIT, P<0.001; Figure 4F ).

Similarly, compared with the nTFHL-AI subgroup, the nTFHL-NOS subgroup displayed relatively lower expression levels of TFH markers, particularly for BCL-6, CD10, and CXCL13. Additionally, the nTFHL-NOS subgroup had a lower median Ki67 index ( Table 4 ).