Polyclonal evolution of lymphoproliferative disorders in XLP1

Signaling lymphocytic activation molecule-associated protein (SAP) deficiency, also known as X-linked lymphoproliferative syndrome type 1 (XLP1), is an inborn error of immunity (IEI) caused by pathogenic variants in SH2D1A. The main clinical features of XLP1 include hemophagocytic lymphohistiocytosis (HLH), hypogammaglobulinemia, and lymphoproliferative disorders (LPDs) (1). Approximately 30% of patients with XLP1 present with LPDs, which are often associated with Epstein-Barr virus (EBV) infection (2). 80% of patients with malignant LPD develop B cell non-Hodgkin lymphoma, predominantly affecting the abdomen and cervical regions. The inability of SAP-deficient T cells to recognize antigen-presenting B cells is one of the reasons for the prevalence of EBV-LPD in patients with XLP1 (3). Immune escape due to an imbalance between T helper 1 (Th1) and Th2 responses has also been proposed as a possible mechanism (4). Somatic variants in lymphoma have also been investigated in other IEIs and have shown a distinct genetic signature. In diffuse large B cell lymphomas (DLBCLs) associated with activated PI3Kδ syndrome, common variable immunodeficiency, and DNA repair disorders, somatic variants in genes, including BRCA2, NCOR1, KLF2, FAS, CCND3, and BRWD3, have been reported at a higher frequency compared with non-IEI DLBCLs (5). IEIs are often complicated by tumors that develop at an early age, typically with a median onset around 20 years (6). Cases of LPD without EBV infection in XLP1 have been reported, and the underlying mechanism of their pathogenesis remains unclear. Therefore, this study aimed to elucidate the mechanism of tumorigenesis through transcriptomic analysis of tumor cells and investigation of somatic variants in patients with XLP1-associated LPDs.

P1 was a male patient with a history of B cell lymphoma at 5 years of age (7). At that time, the lymphoma was considered a primary malignant lymphoma, and since there were no other clinical findings suggestive of XLP, no further investigations for an underlying immunodeficiency were performed. At the age of 18, he developed EBV-HLH, followed by the onset of DLBCL with central nervous system involvement. The patient carried SH2D1A c.208_209insC, p.P70fs*4 variant (Table 1). Pathological examination confirmed DLBCL with a cluster of large EBV-encoded small RNA (EBER)–positive B cells (see Fig. S1 A, Fig. S2 A, and Fig. S3 A).

P2 experienced recurrent infections from infancy and was diagnosed with hypogammaglobulinemia (8). At 4 years of age, genetic analysis revealed an exon 1 deletion in SH2D1A (Table 1). At 5 years of age, the patient presented with left cervical lymphadenopathy. Pathological examination revealed proliferating EBER-positive B cells with preserved follicular structure, suggesting borderline malignancy (see Fig. S1 B, Fig. S2 B, and Fig. S3 B).

P3 was an 8-year-old boy who presented with fever of unknown origin, lymphatic involvement of the liver, spleen, abdominal cavity, lungs, and subcutis, as well as hypogammaglobulinemia. In the liver, T lymphocytic infiltration was observed around the hepatic and portal veins. Genetic analysis revealed an SH2D1A c.162C>A, p.Y54X variant (Table 1). The subcutaneous nodule was diagnosed as T cell lymphoma. An intra-abdominal lymph node biopsy revealed segregation of T cells and EBER-negative B cells, suggesting LPD (see Fig. S1 C, Fig. S2 C, Fig. S3 C, and Fig. S4).

P4 was a 9-year-old boy with a history of recurrent sinusitis who presented with a 2-mo fever, body weight loss, and a sore throat (9). Positron emission tomography (PET) scan revealed increased uptake in the neck, lungs, and intra-abdominal lymph nodes. Biopsy confirmed DLBCL with EBER-positive B cells (see Fig. S1 D, Fig. S2 D, and Fig. S3 D). 1 year later, SH2D1A exons 3–4 deletion was identified (Table 1).

P5 had recurrent pneumonia from age 3, and the IgG level was undetectable with normal B cell counts (9, 10). At age 5, the patient had a persistent EBV infection (78,272 IU/ml in the blood sample) and fever, and a PET scan showed increased uptake in the mediastinum and subcarinal lesions. Pathological examination revealed Hodgkin cells and positive for CD30 and EBER staining, leading to a diagnosis of classical Hodgkin’s lymphoma (nodular sclerosis) (see Fig. S1 E, Fig. S2 E, and Fig. S3 F). The patient carried deletion and insertion variants in SH2D1A (Table 1).

P6 presented with a right cervical mass at 6 years of age. Biopsy confirmed DLBCL with positive EBER staining (see Fig. S1 F, Fig. S2 F, and Fig. S3 F). Fluorescence in situ hybridization revealed a split signal of IRF4. As the brother of the patient also developed lymphoma of colonic origin at 5 years of age, he was tested and found to carry an SH2D1A c.128T>C, p.L43P (Table 1). All six patients underwent hematopoietic cell transplantation (HCT) after chemotherapy and have remained in persistent remission.

A search for somatic variants in LPD samples from P1 to P6 identified 6 non-synonymous variants in the first tumor from P1, 6 in the second tumor from P1, 3 in P4, and 25 in P6 (Table 2). Three variants (CARD11 [two variants] and GNA13) in the first tumor from P1, one variant (MECOM) in the second tumor from P1, and 10 variants (IRF4 [seven variants], P2RY8, KRAS, and CCND3) in P6 were likely pathogenic (Fig. 1). No somatic variants were shared between the first and second tumors in P1.

The first tumor cells from P1 exhibited uniparental disomy (UPD) of chromosome 17q (Fig. 2). No association between 17q UPD and lymphoma has previously been reported. Tumor cells from P6 exhibited UPD of chromosome 12q. This genomic alteration has been observed in follicular lymphoma; however, its role in pathogenesis remains unclear (11).

Tumor RNA samples could not be obtained from P1, P4, and P5. Samples from P2, P3, and P6 were categorized as SAP deficiency-related LPD (SAP-LPD). DLBCL samples from the public data were used as disease controls. An unsupervised clustering analysis was performed between the SAP-LPD and a combined group of c-MYC and BCL2 double expressor (DE)-DLBCL and non–DE-DLBCL (Fig. 3 A). SAP-LPD exhibited a distinct profile compared with DLBCL, suggesting that it forms a separate cluster. Subsequently, Gene Ontology (GO) analysis was performed (Fig. 3 B). In SAP-LPD, pathway analysis revealed a marked downregulation of genes involved in natural killer (NK) cell activation, whereas genes associated with the adaptive immune response were significantly upregulated. Among the differentially expressed genes, various IgH and TCR genes were upregulated in SAP-LPD, in contrast to a more restricted pattern characterized by upregulation of a single dominant gene in DLBCL (Fig. 4). The increased expression of multiple IgH and TCR genes in SAP-LPD supports the notion of polyclonal lymphocyte proliferation (Fig. 5). Representative genes from the top 10 GO terms were extracted, and KLRK1 (NKG2D), STING1, HLA-B, HLA-DQB2, and HLA-E were identified (Fig. 6).

In P1, the first and second tumors harbored entirely distinct somatic variants, indicating independent origins. Within each tumor, the variant allele frequencies were relatively uniform, supporting clonal homogeneity. Several reports of multiple lymphomas arising from distinct clones support multiclonal lymphomagenesis in XLP1 (12, 13). P2 and P3 exhibited nonmalignant LPDs, consistent with the absence of somatic variants. Despite the malignant nature of the LPDs in P4 and P5, formalin-fixed, paraffin-embedded (FFPE) quality, low tumor purity, or a polyclonal lymphoproliferative background may have limited variant detection. In P5, the low tumor cell content, a characteristic feature of Hodgkin lymphoma, may have influenced the analysis. Alternatively, EBV infection, which was detected in both patients, may have contributed to lymphomagenesis. In contrast, P6 was tumorigenic without EBV infection, and detected somatic pathogenic variants likely drove transformation. Notably, CCND3, in which a somatic variant was detected in the P6 tumor, is among the most frequently mutated genes in lymphomas associated with IEIs (5). No recurrent somatic variants unique to XLP1 were detected in this study.

The transcriptome analysis suggests that, unlike conventional DLBCL, which typically arises from monoclonal B cell expansion driven by somatic variants, XLP1-LPD appears to evolve from polyclonal lymphoproliferation. EBV infection under immunosuppression can drive polyclonal B cell proliferation (14, 15). In XLP1, SAP-deficient T cells may fail to recognize EBV-infected B cells, causing reactive polyclonal T cell proliferation (3). Even without EBV infection, intrinsic SAP–T cell defects, such as impaired apoptosis and defective follicular helper T cell function may cause nonspecific B cell activation and expansion (16, 17). In SAP-LPD cases, we observed upregulation of STING1, HLA-B, and HLA-DQB2, suggesting that innate immune activation and enhanced antigen presentation may be common features of the disease. We also observed increased expression of NKG2D, an activating receptor expressed on NK cells and cytotoxic T cells that recognize stress-induced ligands on target cells (18). NKG2D ligand activation promotes lymphocyte proliferation and cytotoxicity and is implicated in autoimmunity (19). In SAP-LPD, this pathway may contribute to sustained lymphocyte activation. In contrast, patients with MAGT1 deficiency, who exhibit reduced NKG2D expression, are more susceptible to EBV infection (20). These findings highlight the importance of tight NKG2D regulation in maintaining immune homeostasis and controlling lymphoproliferation.

In contrast, HLA-E, a nonclassical MHC class I molecule, binds to the inhibitory receptor NKG2A on NK cells and suppresses their cytotoxicity (21). EBV-derived LMP1 upregulates HLA-E expression, thereby promoting immune evasion and the development of LPD (22, 23). GO analysis also revealed downregulation of genes involved in NK cell activation. This paradoxical situation of immune activation with impaired EBV response may reflect SAP deficiency. Therapeutic targeting of the NKG2D or NKG2A axis—already been investigated in oncology and autoimmunity—may offer new strategies for XLP1-LPD (18, 24).

We compared tumor onset age between monogenic immunodeficiency disorders caused by SH2D1A and MAGT1 variants and cancer predisposition syndromes associated with RUNX1, GATA2, and CEBPA variants (Fig. S5). The results highlight the contrast between tumor development driven by impaired T and NK cell function versus tumorigenesis resulting from the intrinsic oncogenic potential of the germline variants. This analysis reveals that tumors develop earlier in immunodeficiencies than in cancer predisposition syndromes. Although HLH remains the most notable prognostic factor for XLP1, lymphoma is also critical (2). Even in non-tumorigenic states, malignant transformation may occur when appropriate triggers are present. Allogenic HCT can be safely performed in asymptomatic patients with XLP1 (25), and early curative HCT is recommended to optimize clinical outcomes.

To our knowledge, this is the first study to identify somatic variants in LPDs in patients with XLP1. However, a major limitation is the small sample size. In addition, in patients with IEIs, it is often difficult to clearly distinguish between lymphoma and LPDs, and in this study, the distinction was made based on histological findings, including preservation of follicular architecture and the presence of monoclonal cell proliferation. Although further validation is required, expression analysis of immune-related molecules such as STING and NKG2D (e.g., by quantitative PCR or immunohistochemistry) in biopsy samples might help clinicians recognize XLP1-LPD. In conclusion, LPD in XLP1 may arise from polyclonal lymphocyte expansion, with tumorigenesis potentially triggered by somatic pathogenic variants. Further studies are warranted to clarify the landscape of XLP1-LPD.

Genetic analysis was performed after obtaining written informed consent from the patients. This study was performed in accordance with the Helsinki declaration and approved by the Ethics Committee of Institute of Science Tokyo (approval number: G2019-004).

Whole-exome sequencing of paired tumor and control DNA samples was performed. Tumor DNA was extracted from tissue sections or FFPE samples and compared with control DNA extracted from peripheral blood mononuclear cells. Library preparation was performed using the SureSelect Human V6 (Agilent Technologies). Sequencing was performed using an Illumina NovaSeq X Plus system (Illumina) with paired-end 100 bp reads. Somatic mutations were identified using the Gnomon pipeline, with sequencing data of non-paired normal tissues used as controls (26). Copy numbers were detected using the CNACS pipeline (27).

Tumor RNA was extracted from the tissue sections. Library preparation was performed using the TruSeq stranded mRNA library (Illumina) following poly A selection. The sequencing was performed on an Illumina NovaSeq X Plus system with paired-end 100 bp reads. Public data on DLBCL (GSE252690) was used. Alignment was performed using Bowtie 2 for RNA-seq data processing (28). Gene expression was quantified using feature Counts. Batch-effect correction was performed using ComBat-Seq, followed by normalization using edgeR. The normalized count data are presented as the trimmed mean of M values normalized counts. Pathway analysis was performed using Metascape (https://metascape.org/gp/index.html#/main/step1) (29).

The supplementary materials include five figures. Fig. S1 shows imaging findings in patients with XLP1 presenting with LPD. Fig. S2 shows hematoxylin and eosin staining of LPD tissues. Fig. S3 shows EBER staining of LPD samples. Fig. S4 shows imaging findings of subcutaneous panniculitis-like T cell lymphoma in P3. Fig. S5 shows a comparison of tumor onset age across different monogenic diseases.

Informed consent was obtained from all patients included in this study and their parents.

Informed consent was obtained from all participants or their parents.

The datasets used in this study are not publicly available to protect participant/patient anonymity. Requests to access the datasets can be made to the corresponding author.

We thank the patients and their families for providing permission to participate in this study.

This study was supported by the Japanese Society of Hematology Research grant and the Lee Kun-hee Child Cancer & Rare Disease Project, Republic of Korea (grant number: 25B-011-0100).

Author contributions: Dan Tomomasa: data curation, formal analysis, investigation, methodology, project administration, software, visualization, and writing—original draft. Akira Nishimura: data curation, formal analysis, investigation, resources, visualization, and writing—original draft, review, and editing. Kenichi Yoshida: data curation, formal analysis, and writing—review and editing. Yui Namikawa: formal analysis and investigation. Doo Ri Kim: investigation. Naoki Sakata: resources and writing—review and editing. Kenichi Sakamoto: resources. Takashi Taga: resources. Yuta Sakai: investigation and resources. Yasuhiro Ikawa: resources and writing—review and editing. Toshiaki Ishida: data curation, resources, visualization, and writing—review and editing. Areum Shin: resources. Keon Hee Yoo: resources and validation. Yae-Jean Kim: funding acquisition and writing—review and editing. Seishi Ogawa: formal analysis and validation. Akihiro Hoshino: visualization and writing—original draft. Tomohiro Morio: supervision and writing—review and editing. Masatoshi Takagi: supervision. Hirokazu Kanegane: conceptualization, funding acquisition, and writing—review and editing.