TP53 and CDKN2A alterations define a poor prognostic subgroup in patients with nodal T follicular helper cell lymphoma

Nodal T follicular helper (TFH) cell lymphoma (nTFHL) is a novel entity introduced in the 5^th edition of the World Health Organization (WHO) Classification of Haematolymphoid Tumors (WHO-HAEM5) [1]. nTFHL (also called as follicular helper T-cell lymphoma in the International Consensus Classification [2, 3]) represents a group of peripheral T-cell lymphoma (PTCL) that expresses TFH immunophenotypic markers, such as PD1, ICOS, CXCL13, CD10, and BCL6, and has TFH-related gene expression signatures [1, 4]. nTFHLs consist of three subtypes, namely nTFHL, angioimmunoblastic type (nTFHL-AI), nTFHL, follicular type (nTFHL-F), and nTFHL, not otherwise specified (nTFHL-NOS). nTFHL-AI, previously designated as angioimmunoblastic T-cell lymphoma, is the prototype of nTFHLs and is characterized by systemic disease, polymorphous lymphoid infiltrates, and extrafollicular follicular dendritic cell (FDC) expansion. In addition, nTFHL-AI often contains B-cell proliferations, which are commonly infected with Epstein–Barr virus (EBV) [1, 5,6,7]. On the other hand, nTFHL-F and nTFHL-NOS, previously designated as follicular T-cell lymphoma and PTCL with TFH phenotype, respectively, have been less characterized in clinical and genetic aspects [1].

Genetic studies have revealed the stepwise accumulation of somatic alterations during nTFHL-AI lymphomagenesis [1, 3]. In early phase, clonal hematopoiesis mutations, such as TET2 and DNMT3A mutations, occur in hematopoietic stem cells [8,9,10,11]. Later, other genetic alterations accumulate in lineage-committed cells which develop into lymphoma. Such alterations include RHOA mutations (with most variants being p.G17V), IDH2 mutations, and mutations in T-cell receptor (TCR) signaling genes, including CD28, VAV1, and PLCG1 [11,12,13,14,15,16,17]. However, whether and how genetic alterations are heterogeneous within nTFHLs and among nTFHL subtypes have not been well investigated. In addition, the genetic differences between nTFHL and peripheral T-cell lymphoma, not otherwise specified (PTCL-NOS), another common PTCL subtype, remain elusive.

Conventionally, most patients with PTCL receive anthracycline-containing regimens such as cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP). However, the prognosis of nTFHL-AI is poor with 5-year overall survival (OS) rate of less than 40% [5,6,7, 18]. Although the combination of brentuximab vedotin and CHP demonstrated superior efficacy against CD30-positive PTCLs, the proportion of CD30 positivity is limited in nTFHL-AI [19, 20]. After achieving remission with chemotherapy, some patients undergo autologous hematopoietic stem cell transplantation (HSCT) as consolidation therapy, which may improve clinical outcomes of nTFHL. In addition, allogeneic HSCT offers a potential cure for a subset of patients, especially those with relapsed or refractory diseases [21, 22]. Historically, the International Prognostic Index (IPI) has been widely used for prognostic prediction in PTCLs [5, 6, 18]. However, whether the integration of genetic information can improve prognostication by clinical and pathological assessment in nTFHL remains unknown.

In this study, we have revealed the entire overview of mutations and copy number alterations (CNAs) in 173 nTFHL patients, identifying four molecular subgroups with distinct clinicopathologic properties. We have also developed a new prognostic scoring system, namely molecular nTFHL prognostic index (mTFHL-PI), incorporating genetic alterations and clinical factors, which can better stratify nTFHL patients.

Detailed methods are available in the Supplementary Information.

Patients

A total of 231 previously untreated PTCL cases, including 173 nTFHL cases (78 nTFHL-AI and 95 nTFHL-NOS cases) and 58 PTCL-NOS cases, at two institutions were enrolled in this study (Supplementary Table S1). nTFHL-F was excluded from this study. All cases were pathologically reviewed and a consensus diagnosis was made by expert hematopathologists (K.C.H., H.M., Y.Matsuno, and K.O.) according to WHO-HAEM5 (particularly the essential diagnostic criteria of nTFHLs) [1]. TFH markers, including PD1, CD10, CXCL13, BCL6, and ICOS, were evaluated, with positivity defined as distinct expression in ≥ 20% of tumor cells as previously described [23], and the TFH phenotype was defined as the presence of at least two TFH markers. Immunohistochemistry (IHC) for CD21, CD23, and/or FDC was performed to evaluate extrafollicular FDC expansion. All samples were collected from patients after obtaining written informed consent, except for previously collected anonymized samples whose secondary use were permitted. This study was approved by the National Cancer Center Institutional Review Board, the Ethical Review Board for Life Science and Medical Research, Hokkaido University Hospital, and the Ethical Committee of Kurume University and was performed in accordance with the Declaration of Helsinki.

Targeted capture sequencing (Targeted-seq)

Genomic DNA was extracted from formalin-fixed paraffin-embedded (FFPE) tumor samples using GeneRead DNA FFPE Kit (QIAGEN, Hilden, Germany). Targeted-seq was performed using a custom SureSelect library (Agilent Technologies, Santa Clara, CA, USA) targeting all coding exons and splice sites of 242 T/NK-cell lymphoma-associated genes (Supplementary Table S2). Sequence alignment, mutation calling, and CNA detection were performed using Genomon pipeline (RRID:SCR_022989) version 2.6.3 and CNACS (RRID:SCR_022988), as previously described [24, 25]. Focal deletions affecting TP53 and CDKN2A and arm-level CNAs were analyzed.

Survival analysis

Survival data were available for 152 nTFHL and 55 PTCL-NOS cases. Of 138 cases whose first-line therapy information was available, 129 (93%) received CHOP or CHOP-like chemotherapy (Supplementary Table S3). No case received brentuximab vedotin as a part of first-line therapy. OS was calculated from the time of diagnosis, and observations were censored at the last follow-up. At the last follow-up, 85 nTFHL and 17 PTCL-NOS cases were still alive and their median follow-up periods were 43 and 27 months, respectively. Univariable and multivariable survival analyses were performed according to the Kaplan–Meier method with log-rank test and Cox proportional hazards model method incorporating molecular subgroup or genetic alteration status and clinical factors, using the R package survival (RRID:SCR_021137) version 3.5–5.

Landscape of driver alterations in nTFHL

We performed targeted-seq in 173 untreated nTFHL cases, with a mean depth of ×ばつ (Supplementary Fig. S1A; Supplementary Tables S1 and S2). A total of 805 somatic mutations, including 621 single nucleotide variants (SNVs), 5 multi-nucleotide variants (MNVs), and 179 insertion-deletions (indels), were identified, with a median of 4 (range, 0–16) mutations per case (Supplementary Table S4). A total of 25 genes were significantly mutated (in ≥ 3 cases and Q < 0.25) (Supplementary Table S5) [26]. Besides them, among known drivers in nTFHL [11,12,13,14,15,16,17, 24, 27], ten additional genes exhibited mutations in ≥ 3 cases. Among the 35 driver genes, 8 genes were affected in more than 5% of cases. In accordance with previous studies [11,12,13,14,15,16,17, 24, 27], TET2 was most frequently mutated (67%), followed by RHOA (46%), IDH2 (23%), DNMT3A (18%), PLCG1 (9%), TP53 (8%), HLA-A (5%), and HLA-B (5%) (Fig. 1A).

We identified four driver genes which have not been reported to be recurrently mutated in nTFHL, namely TET3 (3%), HLA-C (3%), NRAS (2%), and KLF2 (2%) (Fig. 1A; Supplementary Fig. S1B). TET3, belonging to the ten-eleven translocation (TET) family of methylcytosine dioxygenases [28], was recurrently affected by loss-of-function mutations and invariably co-occurred with TET2 mutations (Fig. 1B). In addition, there were recurrent inactivating mutations in HLA-C and KLF2, which have been reported in B-cell lymphomas [29, 30]. On the other hand, activating hotspot mutations in NRAS (particularly involving G12) were detected. Although recurrent STAT3 mutations have been reported in nTFHLs, we identified a novel mutational hotspot at D570, in addition to the previously reported ones in nTFHL and other PTCLs (Supplementary Fig. S1B) [14, 17, 24, 31].

Copy number analysis detected 255 arm-level CNAs (range, 0–20 per case), including 148 gains and 107 losses (Supplementary Fig. S2A). TP53 was affected by both arm-level 17p deletions (n = 7) and focal deletions (n = 4) (Fig. 1C; Supplementary Table S6). When mutation and CNA were combined, 14 (8%) cases harbored TP53 alterations, which were predominantly biallelic (Supplementary Fig. S2B, C). Furthermore, 10 (6%) cases had focal deletions of CDKN2A, including homozygous deletions (Fig. 1C; Supplementary Fig. S2D). Overall, 145 (84%) and 16 (9%) cases carried at least one driver mutation and CNA, respectively (Fig. 1B). When considering deletions affecting TP53 and CDKN2A, 145 (84%) cases harbored at least one driver alteration. These findings suggest considerable genetic heterogeneity in nTFHL subtypes.

Molecular subgroups harboring characteristic genetic alterations in nTFHL

To identify tumors with shared genetic features, we applied unsupervised hierarchical clustering to driver alterations (observed in > 5 cases: 18 driver mutations and TP53 and CDKN2A deletions) and 16 arm-level CNAs (observed in > 5 cases). We identified four distinct clusters with characteristic genetic alterations based on the dendrogram (Supplementary Fig. S3). To develop clinically applicable classification, we created a simple decision tree, which classified nTFHL cases into four molecular subgroups, based on the results of hierarchical clustering (Fig. 2A). The molecular subgroups included the TR-I (+) (n = 39) and TR-I (−) (n = 73) subgroups based on TET2 and/or RHOA mutations with and without IDH2 mutations; the AC53 subgroup (n = 19) based on aneuploidy with TP53 and/or CDKN2A alterations; and the NSD subgroup (n = 42) with no subgroup-defining alterations (namely without any of TET2, RHOA, IDH2, TP53, and CDKN2A alterations) (Figs. 1B and 2B). NSD had significantly fewer driver mutations than other subgroups, although the sequencing depth was comparable across subgroups (Fig. 2C; Supplementary Fig. S4A). By contrast, AC53 showed a much larger number of CNAs, particularly copy number losses (Fig. 2C; Supplementary Fig. S4B, C). Although many arm-level CNAs were most frequent in AC53, 5p and 5q gains were enriched in TR-I (+) (Supplementary Fig. S4D), consistent with previous findings that these gains are characteristic of nTFHL-AI [32].

With regard to functional pathways, somatic alterations in TFH-related genes (TET2, RHOA, IDH2, and DNMT3A) were enriched in TR-I (+) and TR-I (−), while those affecting DNA repair genes (TP53 and CDKN2A) were exclusively observed in AC53, reflecting their subgroup-defining potential (Fig. 2D). While the frequency of alterations in TCR signaling/NF-κB pathway genes (including CD28, PLCG1, and VAV1) was comparable across TR-I (+), TR-I (−), and AC53, it was significantly lower in NSD.

Clinical and pathological differences across molecular subgroups in nTFHL

Then, we investigated the association between molecular subgroups and pathological and clinical findings (Fig. 3A). Each of the four molecular subgroups included both nTFHL-AI and nTFHL-NOS cases, with significant enrichment of nTFHL-AI in TR-I (+) and nTFHL-NOS in NSD. The number of positive TFH markers was highest in TR-I (+), while NSD had the lowest. In terms of individual TFH markers, the proportion of PD1 and ICOS positivity was comparable across all subgroups. However, TR-I (+) showed a higher proportion of CD10, BCL6, and CXCL13 positivity than other subgroups (Supplementary Fig. S5A). Particularly, CD10-positive cases were rarely observed in AC53 and NSD, suggesting high specificity of this marker. Therefore, the TR-I (+) cases possess typical features of nTFHL-AI, such as IDH2 mutations and a broad range of TFH marker positivity [15, 16, 27].

Clinically, TR-I (+) was associated with advanced-stage disease, consistent with the typical clinical features of nTFHL-AI (Fig. 3A) [5,6,7, 18]. On the other hand, AC53 showed male predominance and more extensive extranodal involvement. Notably, NSD was associated with younger age, limited-stage disease, better performance status, and lower IPI score, suggesting a more favorable nature of this subgroup. Lactate dehydrogenase (LDH) levels and bone marrow involvement status were almost comparable across molecular subgroups (Supplementary Fig. S5B). Taken together, these observations suggest diverse clinical and pathological features across the molecular subgroups in nTFHL.

Prognostic impact of molecular subgroups in nTFHL

We assessed the impact of molecular subgroups on clinical outcomes in 152 nTFHL cases, for whom survival data were available, and observed substantially different survival rates across molecular subgroups in univariable analysis (Fig. 3B). AC53 showed the worst prognosis among the molecular subgroups, with a median OS of 15 months. By contrast, NSD was significantly associated with better OS, with a median OS of 181 months. The survival rates of TR-I (+) and TR-I (−) were nearly identical, with a median OS of 61 months, and between that of AC53 and NSD. Thus, the two subgroups were combined in subsequent survival analyses. Similar results were obtained when only 109 patients who did not undergo HSCT were assessed (Supplementary Fig. S5C).

We also investigated the prognostic impact of individual genetic alterations (observed in ≥ 10 cases). In univariable analysis, TP53 and CDKN2A alterations were significantly associated with shorter OS (P = 1.9 ×ばつ 10⁻⁶ and 0.00019, respectively), while there were no other significant alterations (Supplementary Fig. S5D, E ; Supplementary Table S7). Accordingly, cases harboring TP53 and/or CDKN2A alterations showed a significantly worse prognosis, with a median OS of 15 months (P = 5.0 ×ばつ 10⁻⁷) (Supplementary Fig. S5F). These results suggest that molecular classification can predict patient outcomes in nTFHL. Furthermore, TP53 and CDKN2A alterations are considered to account for the adverse prognostic impact of AC53.

Influence of the presence/absence of driver alterations in the NSD subgroup

Then, we investigated the genetic properties of NSD, showing the best prognosis in nTFHL. The spectrum of driver alterations was different between NSD and other subgroups, suggesting the unique molecular features of NSD. Specifically, genetic alterations associated with immune surveillance, epigenetic regulation, and transcriptional regulation were enriched in NSD (Fig. 4A). Importantly, 28 (67%) cases lacked driver alterations in NSD (Fig. 1B).

To evaluate clonality in NSD, we quantified TCR clonotypes using targeted-seq data. No preferential V or J segment usage for the most abundant TCR clonotype was observed in nTFHL (Supplementary Fig. S6A, B). The fraction of the most abundant TCR clonotype was significantly correlated with the median allele frequency of driver mutations (Pearson’s r = 0.58, P = 3.1 ×ばつ 10⁻¹² for TRAD; Pearson’s r = 0.53, P = 5.0 ×ばつ 10⁻⁸ for TRB; Fig. 4B). The clone fraction in AC53 was larger than in other subgroups (Fig. 4C). By contrast, the clone fraction of NSD was the smallest, consistent with the smaller number of driver mutations in NSD (Fig. 2C). However, NSD cases without driver alterations showed a similar clone fraction to those with driver alterations (Fig. 4D), suggesting that clonal T-cell proliferation occurs without known driver alterations in these cases.

More importantly, the presence of driver alterations was significantly associated with a worse prognosis in NSD (P = 0.014), although the pathological and clinical characteristics were comparable between those with and without driver alterations (Fig. 4E; Supplementary Fig. S7A, B). Indeed, NSD cases without driver alterations exhibited typical pathological features of nTFHL-AI or nTFHL-NOS (Supplementary Fig. S7B). For NSD cases without driver alterations, the median OS was not reached, and the 5-year OS rate was 88%. By contrast, the NSD cases with driver alterations showed a comparable prognosis to the TR-I (+/−) cases. Therefore, the presence/absence of driver alterations is a critical determinant of clinical outcome in nTFHL and the absence of driver alterations primarily accounts for the positive prognostic impact of NSD.

Association of EBV status with genetic and clinical phenotypes in nTFHL

As EBV infection is frequently observed in B cells from nTFHL-AI [1, 5,6,7], we evaluated EBV status by targeted-seq. Overall, 147 (85%) of the 173 nTFHL cases were positive for EBV, including 123 weak positive (+) and 24 strong positive (++) cases (Fig. 5A). Overall, nTFHL showed significantly stronger positivity than 11 samples of reactive lymph nodes, of which 3 samples were weak positive for EBV. Then, we performed CD20 staining and EBV-encoded small RNA in situ hybridization (EBER-ISH) in 96 nTFHL cases. Our analysis revealed a significant positive correlation between the ratio of EBV-aligned reads to total reads and the number of EBV-positive B cells (Fig. 5B), supporting the validity of our next-generation sequencing (NGS)-based EBV detection. The proportion of EBV-strong positive cases tended to be higher in TR-I (+) and TR-I (−) than in other subgroups (Supplementary Fig. S8A). Regarding genetic alterations, the number of driver mutations and CNAs was comparable across EBV-negative, -weak positive, and -strong positive cases. Although not statistically significant, the frequency of TFH-related mutations positively correlated with the extent of EBV positivity (Supplementary Fig. S8B, C). EBV-strong positive cases showed greater LDH levels and higher IPI score as well as a larger clone fraction (Fig. 5C, D; Supplementary Fig. S8D, E). Furthermore, univariable analysis demonstrated that EBV-strong positive cases exhibited a worse prognosis than EBV-weak positive and EBV-negative cases, with a median OS of 16 months (P = 0.0049 and 0.0055, respectively), although similar survival outcomes were observed between EBV-weak positive and EBV-negative cases with a median OS of 61 months and not reached respectively (Fig. 5E). These results suggest that EBV strong positivity is weakly associated with TFH-related alterations and confers more aggressive clinical behavior.

Comparison between nTFHL-AI, nTFHL-NOS, and PTCL-NOS

We then separated nTFHL cases into nTFHL-AI and nTFHL-NOS and compared genetic and clinical profiles between nTFHL-AI, nTFHL-NOS, and PTCL-NOS. To address this, we performed targeted-seq in 58 cases with PTCL-NOS, with a mean depth of ×ばつ (Supplementary Fig. S9A). A total of 233 somatic mutations, including 190 SNVs, 3 MNVs, and 40 indels, were identified, with a median of 4 per case (0–19 mutations per case) (Supplementary Table S8). Consistent with previous reports [24, 32], TET2 was most frequently mutated (33%), followed by DNMT3A (16%), RHOA (16%), TP53 (10%), PLCG1 (7%), and CARD11 (7%) in PTCL-NOS (Supplementary Fig. S9B). In addition, TP53 and CDKN2A deletions were observed in 10% and 10% of cases, respectively (Supplementary Fig. S9C; Supplementary Table S9). Overall, 34 (59%) and 10 (17%) cases carried at least one driver mutation and CNA, respectively (Supplementary Fig. S9D). When considering deletions affecting TP53 and CDKN2A, 36 (62%) cases harbored at least one driver alteration.

Regarding genetic alterations, the number of CNAs was comparable between nTFHL-AI, nTFHL-NOS, and PTCL-NOS, whereas PTCL-NOS cases showed fewer driver mutations with lower allele frequencies than nTFHL, in which more driver mutations were found in nTFHL-AI than in nTFHL-NOS (Fig. 6A; Supplementary Fig. S10A, B). While TP53 and CDKN2A alterations were or tended to be more common in PTCL-NOS, the frequencies of TET2, RHOA, and IDH2 mutations were by far highest in nTFHL-AI (Fig. 6B). The mutational frequency of nTFHL-NOS was intermediate between those of nTFHL-AI and PTCL-NOS. In addition, almost all mutations affected G17 in RHOA and R172 in IDH2 in nTFHL, whereas missense mutations in other positions were frequently observed in PTCL-NOS (Supplementary Fig. S10C). Reflecting these differences, somatic alterations in TFH-related genes were overrepresented in nTFHL-AI (Fig. 6C; Supplementary Fig. S10D). By contrast, those related to DNA repair were more common in PTCL-NOS. We then classified PTCL-NOS cases into molecular subgroups according to the same decision tree as nTFHL. In PTCL-NOS, 17, 13, and 28 cases were categorized into the TR-I (−), AC53, and NSD subgroups, respectively (Fig. 6D). Among them, the proportion of TR-I (+) was higher in nTFHL-AI than in nTFHL-NOS and PTCL-NOS, whereas the proportion of TR-I (−) was comparable across nTFHL-AI, nTFHL-NOS, and PTCL-NOS. On the other hand, AC53 and NSD were most frequent in PTCL-NOS.

As for clinical factors, almost no significant differences were observed between nTFHL-AI, nTFHL-NOS, and PTCL-NOS, although nTFHL-AI was associated with older age and more advanced stage (Supplementary Fig. S10E). Next, we assessed the influence of the histological subtype on clinical outcomes. Consistent with a previous report [33], nTFHL-AI and nTFHL-NOS showed almost identical survival curves, which were significantly better than PTCL-NOS (Fig. 6E). Therefore, we combined nTFHL-AI and nTFHL-NOS cases and compared their survival with that of PTCL-NOS in each molecular subgroup. Remarkably, nTFHL cases showed better survival than PTCL-NOS cases in TR-I (+/−) and NSD (P = 1.1 ×ばつ 10⁻⁵ and 0.00076, respectively) (Fig. 6F). By contrast, nTFHL and PTCL-NOS had similar survival rates in AC53. These observations suggest that AC53 represents a distinct entity irrespective of histological subtype, while pathological evaluation is important when TP53 and CDKN2A alterations are absent.

Multivariable risk stratification of nTFHL patients

We then evaluated the relative effects of molecular subgroup using Cox proportional hazards modeling, with IPI and EBV positivity as a covariate, in which we incorporated the presence/absence of driver alterations instead of NSD based on the above results (Supplementary Table S3). As a result, AC53, the presence of driver alterations, and IPI high-risk were independently associated with a worse prognosis (Fig. 7A). Therefore, genomic profiling can improve prognostic prediction independently of clinical factors in nTFHL.

Based on the relative risks of these variables, we developed mTFHL-PI by assigning one point for each of the following: AC53 (TP53 and/or CDKN2A alterations), the presence of driver alterations, and IPI high-risk. According to mTFHL-PI, 23 (16%), 71 (49%), 44 (30%), and 7 (5%) cases had a score of 0, 1, 2, and 3, respectively (Fig. 7B). Then, nTFHL cases were classified into three risk categories: low (mTFHL-PI = 0), intermediate (1), and high (≥2). These categories showed significantly different prognoses, with a median OS of 181, 67, and 20 months, respectively (P = 3.1 ×ばつ 10⁻⁷) (Fig. 7C). Similar results were obtained when only 104 patients who did not undergo HSCT were assessed (Supplementary Fig. S11). Taken together, these observations suggest that mTFHL-PI is prognostically informative in nTFHL.

Through clinicogenomic analyses in the largest cohort of nTFHL patients, we have delineated the entire picture of somatic alterations. We identified four molecular subgroups with distinct genetic and clinical features, namely TR-I (+), TR-I (−), AC53, and NSD, suggesting further genetic heterogeneity of nTFHL subtypes. We also characterized the genetic differences between nTFHL-AI, nTFHL-NOS, and PTCL-NOS, demonstrating that nTFHL-NOS harbors genetic profiles intermediate between nTFHL-AI and PTCL-NOS. Remarkably, nTFHL shows better survival than PTCL-NOS in TR-I (+/−) and NSD, suggesting not only the prognostic difference between nTFHL and PTCL-NOS but also the complementary utility of pathological assessment and genomic profiling. Finally, we propose a new prognostic scoring system (mTFHL-PI), which can successfully stratify nTFHL patients using AC53, the presence of driver alterations, and IPI high-risk.

Among the four molecular subgroups, AC53 exhibited the worst prognosis in nTFHL, similar to that in PTCL-NOS [24, 32]. Notably, AC53 barely carried clonal hematopoiesis mutations in both nTFHL and PTCL-NOS, highlighting the difference in pathogenesis from TR-I (+/−). Furthermore, AC53 showed a characteristic immunophenotype with almost no CD10 positivity, in sharp contrast to TR-I (+/−). Thus, AC53 is considered to represent a distinct entity regardless of histological subtype. This notion is supported by a recent study in which TP53 mutations and deletions were associated with inferior survival in various PTCL subtypes treated with CHOP-based chemotherapy [34]. Although the role of autologous and allogeneic HSCT remains controversial in PTCLs [21, 22], the AC53 patients would potentially benefit from more intensive treatment, including HSCT.

TR-I (+) and TR-I (−) share clinical and genetic features, although TR-I (+) contains more typical nTFHL-AI cases, consistent with previous reports [15, 16]. In mouse models, Tet2 loss combined with Rhoa p.G17V or Idh2 p.R172K expression in CD4⁺ T cells induces T-cell lymphomas recapitulating nTFHL-AI through deregulating transcriptomic and/or epigenetic profiles of TFH cells [35,36,37]. Of note, recent clinical trials have shown that epigenetic modifiers, including histone deacetylase inhibitors and hypomethylating agents, are effective against PTCLs and their response is associated with TFH phenotype [38,39,40,41,42,43,44,45]. As substantial genetic heterogeneity exists in nTFHL, genomic profiling could potentially improve prediction of response to epigenetic modifiers. Further studies investigating the efficacy of epigenetic modifiers, in genetically defined TR-I (+/−) tumors would be warranted.

Another notable finding in this study is a strong prognostic impact of the presence/absence of driver alterations in nTFHL. Within NSD, tumors without driver alterations showed a comparable clone fraction with those with driver alterations. However, given that recent single-cell analyses have revealed that clonal T-cell expansion is widely observed in various pathological conditions [46,47,48], these tumors may include those with reactive lymphoid proliferation that morphologically mimics nTFHL. Thus, these observations suggest that evaluating the presence/absence of driver alterations using NGS would be more useful than TCR clonality analysis to identify a subset of patients who are not suitable for standard chemotherapy. In addition, less toxic treatments or even watchful waiting can be more favorable for this subset. Taken together, integrating NGS-based genomic profiling would support the differential diagnosis and management of these conditions.

In several B-cell and T/NK-cell lymphoma subtypes, such as Burkitt lymphoma and extranodal NK/T-cell lymphoma, EBV infection occurs in tumor cells and is involved in their transformation [49]. By contrast, EBV infection is observed in non-tumor cells (predominantly in B cells) in nTFHL. In consistent with previous reports using EBER-ISH [5,6,7], approximately 85% of nTFHL cases were EBV-positive based on NGS methods in our study, although NGS cannot determine which cell type is infected. Among them, 14% of cases were strongly positive for EBV, which was associated with aggressive phenotype. These findings imply that EBV-infected B cells affect tumor characteristics in the microenvironment. This hypothesis is supported by a recent finding that clonal hematopoiesis-derived B cells function as a niche for T-cell lymphomagenesis in a nTFHL-AI mouse model [50]. A retrospective study showed that the addition of anti-CD20 monoclonal antibody rituximab to CHOP/CHOP-like regimens improves overall response for nTFHL-AI patients, although their survival does not change [51]. Therefore, further studies investigating the role and indication for therapeutic strategy targeting EBV-infected B cells in nTFHL are warranted.

In summary, we have illustrated the genetic landscape of nTFHLs and defined four molecular subgroups with biological and clinical relevance. Our findings provide potential implications for better prognostication and the development of new treatment strategies for nTFHLs.

The data generated in this study are available upon request from the corresponding author.

The authors thank the hematologists and pathologists who were involved in North Japan Hematology Study Group (Supplementary Appendix) for patient care and Y.Hokama, F.Ueki, and Yoshiko Ito for technical assistance. The supercomputing resources were provided by the Human Genome Center, the Institute of Medical Science, The University of Tokyo.

This work was supported by the Japan Society for the Promotion of Science (JSPS) KAKENHI (JP21H05051), Japan Science and Technology Agency Moonshot R&D Program (JPMJMS2022), and Takeda Science Foundation to K.Kataoka, and by JSPS KAKENHI (JP21H02775) to M.N.

1. These authors contributed equally: Yuta Ito, Joji Shimono, Keisuke Kawamoto.

Authors and Affiliations

1. Division of Molecular Oncology, National Cancer Center Research Institute, Tokyo, Japan

   Yuta Ito, Yasunori Kogure, Mariko Tabata, Yuki Saito, Kota Mizuno, Sara Horie, Yosuke Mizukami, Junji Koya, Koichi Murakami & Keisuke Kataoka

2. Division of Clinical Oncology and Hematology, Department of Internal Medicine, The Jikei University School of Medicine, Tokyo, Japan

   Yuta Ito

3. Department of Hematology, Hokkaido University Faculty of Medicine, Sapporo, Japan

   Joji Shimono, Takanori Teshima & Masao Nakagawa

4. Department of Pathology, Kurume University School of Medicine, Kurume, Japan

   Keisuke Kawamoto, Hiroaki Miyoshi & Koichi Ohshima

5. Institute of Pathology, Julius-Maximilians-Universität Würzburg, Würzburg, Germany

   Keisuke Kawamoto

6. Center for Development of Advanced Diagnostics, Hokkaido University Hospital, Sapporo, Japan

   Kanako C. Hatanaka

7. Department of Urology, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan

   Mariko Tabata

8. Department of Gastroenterology, Keio University School of Medicine, Tokyo, Japan

   Yuki Saito, Sara Horie & Yosuke Mizukami

9. Division of Hematology, Department of Medicine, Keio University School of Medicine, Tokyo, Japan

   Kota Mizuno, Koichi Murakami & Keisuke Kataoka

10. Research Division of Genome Companion Diagnostics, Hokkaido University Hospital, Sapporo, Japan

    Yutaka Hatanaka

11. Division of Genome Analysis Platform Development, National Cancer Center Research Institute, Tokyo, Japan

    Kenichi Chiba, Ai Okada & Yuichi Shiraishi

12. Department of Surgical Pathology, Hokkaido University Hospital, Sapporo, Japan

    Yoshihiro Matsuno

Contributions

Y.I.: Formal analysis, investigation, data curation, writing – original draft, writing – review and editing, visualization. J.S.: Investigation, data curation, resources, writing – review and editing. K.Kawamoto: Investigation, resources, writing – original draft, writing – review and editing. K.C.H.: Investigation, resources, writing – review and editing. Y.K.: Formal analysis, writing – original draft, writing – review and editing. M.T.: Formal analysis, writing – review and editing. Y.Saito.: Formal analysis, writing – review and editing. K.Mizuno: Formal analysis, writing – review and editing. S.H.: Formal analysis, writing – review and editing. Y.Mizukami: Formal analysis, writing – review and editing. J.K.: Investigation, writing – review and editing. K.Murakami: Investigation, writing – review and editing. T.T.: Investigation, resources, writing – review and editing. Y.H.: Investigation, resources, writing – review and editing. K.C.: Software, writing – review and editing. A.O.: Software, writing – review and editing. Y.Shiraishi: Software, writing – review and editing. H.M.: Investigation, resources, writing – original draft, writing – review and editing. Y.Matsuno: Investigation, resources, writing – review and editing. K.O.: Conceptualization, investigation, resources, writing – original draft, writing – review and editing, project administration. K.Kataoka: Conceptualization, formal analysis, writing – original draft, writing–review and editing, project administration, funding acquisition. M.N.: Conceptualization, resources, data curation, writing – original draft, writing – review and editing, project administration, funding acquisition.

Corresponding authors

Correspondence to Koichi Ohshima, Keisuke Kataoka or Masao Nakagawa.

Competing interests

K.C.H. has received research funding from NEC Corporation, Eli Lilly, and SEKISUI CHEMICAL. Y.K. has received honoraria from Kyowa Kirin, Nippon Shinyaku, Daiichi Sankyo, and Takeda Pharmaceutical. J.K. has received honoraria from TOMY DIGITAL BIOLOGY, Scrum, and Eisai. T.T. has received honoraria from Astellas Pharma, AstraZeneca, Kyowa-Kirin, SymBio Pharmaceuticals, Sumitomo Pharma, Nippon Shinyaku, Novartis, Janssen Pharmaceutical, and Bristol-Myers Squibb; has received research grants from Asahi Kasei Pharma, Eisai, Ono Pharmaceutical, Kyowa-Kirin, Shionogi, Sumitomo Pharma, Chugai Pharmaceutical, and Nippon Shinyaku; and has received research funding from Astellas Pharma, Otsuka Pharmaceutical, LUCA Science, and Priothera SAS. Y.H. has received honoraria from Eli Lilly, Daiichi Sankyo, AstraZeneca, Merck, Novartis, and Merck Sharp & Dohme; has received research funding from NEC Corporation, Eli Lilly, Shionogi, Daiichi Sankyo, and Sysmex; and has received consultancy fee from NEC Corporation. K.Kataoka has received honoraria from Ono Pharmaceutical, Eisai, Astellas Pharma, Novartis, Chugai Pharmaceutical, AstraZeneca, Sumitomo Pharma, Kyowa Kirin, Janssen Pharmaceutical, Takeda Pharmaceutical, Otsuka Pharmaceutical, SymBio Pharmaceuticals, Bristol Myers Squibb, Pfizer, Nippon Shinyaku, Daiichi Sankyo, Alexion Pharmaceuticals, AbbVie, Meiji Seika Pharma, Sanofi, Sysmex, Mundipharma, Incyte Corporation, and Kyorin Pharmaceutical; has received research funding from Asahi Kasei Pharma, Eisai, Otsuka Pharmaceutical, Ono Pharmaceutical, Kyowa Kirin, Shionogi, Takeda Pharmaceutical, Sumitomo Dainippon Pharma, Chugai Pharmaceutical, Teijin Pharma, Japan Blood Products Organization, Mochida Pharmaceutical, JCR Pharmaceuticals, Nippon Shinyaku, Chordia Therapeutics, and Meiji Seika Pharma; holds individual stocks in Asahi Genomics; and has a patent for Genetic alterations as a biomarker in T-cell lymphomas licensed to Kyoto University and PD-L1 abnormalities as a predictive biomarker for immune checkpoint blockade therapy licensed to Kyoto University. M.N. has received honoraria from Takeda Pharmaceutical, Meiji Seika Pharma, AstraZeneca, and Mundipharma; and has received research funding from Takeda Pharmaceutical and AbbVie. All remaining authors have declared no conflicts of interest.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions