Cold Agglutinin Disease–Like Genetic Signatures in Lymphoplasmacytic Lymphoma–Like Cases Challenge the WHO Fifth Edition Classification Authors: Yuki Sato, Daisuke Koyama, Shoki Yamada, Kengo Suzuki, Masahiko Fukatsu, Yuka Oka, Yuko Hashimoto, Takayuki Ikezoe Published: 2026 ABSTRACT ABSTRACTBackgroundThe fifth edition of the World Health Organization (WHO) classification defines cold agglutinin disease (CAD) as a clonal B‐cell lymphoproliferative disorder occurring in the absence of overt lymphoma, whereas cases associated with lymphoma are classified as cold agglutinin syndrome (CAS). However, whether this distinction fully reflects the biological spectrum of cold agglutinin–associated disorders remains uncertain.MethodsWe performed a comprehensive pathological review, immunoglobulin gene analysis and whole‐exome sequencing in eight patients with clinically well‐characterised CAD or CAS.ResultsFive cases showed pathological findings typical of CAD, whereas two cases demonstrated lymphoma‐like bone marrow infiltrates resembling lymphoplasmacytic lymphoma (LPL). Nevertheless, no CD5‐positive or CD10‐positive clonal B‐cell population suggestive of a specific lymphoma subtype was identified by flow cytometry. Six patients showed highly restricted IGHV4‐34 usage, and mutations involving KMT2D and CARD11 were identified. No pathogenic MYD88 L265P or CXCR4 mutations were detected in any case. Notably, the two morphologically LPL‐like cases lacked the canonical genetic hallmarks of LPL and instead exhibited immunogenetic and mutational features characteristic of CAD.ConclusionThese findings demonstrate that CAD‐like genetic signatures can be present even in cases with lymphoma‐like morphology that would currently be classified as CAS under the WHO criteria. Our results highlight a clinically relevant genotype–phenotype discordance and support the integration of molecular hallmarks into diagnostic algorithms to better reflect the biological spectrum of cold agglutinin–associated disorders. Trial Registration: The authors have confirmed clinical trial registration is not needed for this submissionKeywords: autoimmune haemolytic anaemia, cold agglutinin disease, cold agglutinin syndrome, lymphoplasmacytic lymphoma, WHO classification Introduction Cold agglutinin disease (CAD) is a distinct clonal B‐cell lymphoproliferative disorder characterised by complement‐mediated autoimmune haemolytic anaemia driven by monoclonal cold‐reactive IgM [1]. In the fifth edition of the World Health Organization (WHO) classification and the International Consensus Classification (ICC), CAD was defined for the first time as a neoplastic entity occurring in the absence of overt lymphoma, whereas cases associated with lymphoid malignancies are designated cold agglutinin syndrome (CAS) [2, 3]. This conceptual separation reflects advances in the molecular characterisation of CAD, which has revealed a highly restricted immunoglobulin gene repertoire, most notably preferential IGHV4‐34 usage. These studies have also defined a recurrent mutational landscape dominated by KMT2D and CARD11 alterations in the absence of MYD88 L265P and CXCR4 mutations that typify lymphoplasmacytic lymphoma (LPL) [1, 4, 5, 6, 7, 8]. Furthermore, IGLL5, a well‐established target of activation‐induced cytidine deaminase‐mediated off‐target somatic hypermutation, is frequently mutated across a broad spectrum of mature B‐cell neoplasms, including CAD [9, 10]. The presence of IGLL5 mutations in CAD is biologically plausible given the extensive somatic hypermutation and germinal centre (GC)‐experienced B‐cell origin of this disorder, in contrast to LPL, which is generally considered to arise from post‐GC B cells and does not depend on classical GC reactions [11, 12]. The cold agglutinins in CAD are encoded almost exclusively by the IGHV4‐34 gene segment, in combination with either IGKV3‐20 or IGKV3‐15 [1]. Framework region (FR) 1 of IGHV4‐34 is specific for the I antigen on the surface of erythrocytes, providing a mechanistic explanation for the markedly restricted immunoglobulin gene usage observed in CAD [1, 13]. These insights have helped establish CAD as a biologically distinct bone marrow‐based lymphoproliferative disorder [4, 14, 15, 16]. Nevertheless, a subset of cold agglutinin–associated cases encountered in routine practice remains difficult to classify according to current criteria [17, 18, 19]. Some patients have bone marrow infiltrates that fulfil the morphological thresholds for low‐grade B‐cell lymphoma, particularly LPL, yet lack the genetic hallmarks of these entities and instead display features characteristic of CAD [20]. Under the current WHO framework, such cases are inevitably classified as CAS, although whether this categorisation accurately reflects their underlying biology remains uncertain. Systematic molecular and pathological analyses of such borderline cases are scarce. In this study, we performed a comprehensive pathological review, immunoglobulin gene analysis and whole‐exome sequencing in eight patients with clinically well‐documented CAD or CAS. Our aim was to investigate the biological basis of diagnostically challenging cases and to explore whether genetic profiling can refine disease classification within the current WHO framework. Methods Patients We collected data from eight patients who were diagnosed with clinically well‐documented cold agglutinin–associated disorders at our institution between October 2021 and September 2025. This study was approved by the Institutional Review Board of Fukushima Medical University, and whole‐exome sequencing data were deposited in the Japanese Genotype–phenotype Archive (JGA). All participants provided written informed consent. All patients had a clinical history of CAD with varying degrees of anaemia, a positive C3d‐specific direct antiglobulin test, and a cold agglutinin titre ≥ 1:64. To evaluate for concomitant lymphoma, flow cytometric analysis was performed using bone marrow aspirate samples obtained at diagnosis at an external laboratory (BML, Tokyo, Japan). The flow cytometry findings are given in Table S1. Neither lymphadenopathy nor splenomegaly was detectable on computed tomography in any of the patients [2]. Pathological Evaluation Formalin‐fixed, paraffin‐embedded bone marrow aspirate clot sections obtained at diagnosis were independently reviewed by board‐certified haematopathologists. Morphological assessment focused on the pattern and extent of B‐cell infiltration, plasmacytic differentiation and associated histological features. DNA Extraction Bone marrow mononuclear cells were isolated from frozen aspirate samples using Lympholyte‐H (Cedarlane, Burlington, USA). Genomic DNA was extracted using the QIAamp DNA Blood Kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions. DNA was also extracted from oral mucosa samples using MonoFas gDNA buccal swabs Extraction Kit VIII (ANIMOS, Saitama, Japan) and was used as matched normal controls. Immunoglobulin Heavy or Light Chain Gene Sequencing Rearranged immunoglobulin heavy chain genes were analysed using the IGH Somatic Hypermutation Assay v2.0 (Invivoscribe, San Diego, CA, USA). Immunoglobulin light chain gene rearrangements were amplified using BIOMED‐2‐based in‐house protocols [21, 22]. The purified PCR products were subsequently subjected to bidirectional sequencing using the corresponding assay primers [23]. Whole‐Exome Library Preparation and Sequencing Whole‐exome libraries were prepared with the Agilent SureSelect Human All Exon V6‐Post Kit (Agilent Technologies, Santa Clara, USA) and sequenced with 2 × 150 bp paired‐end reads on the Illumina NovaSeq 6000 or X platform [24]. Raw sequencing reads were assessed using FastQC. Reads were aligned to the GRCh38 reference genome using BWA‐MEM (v0.7.17), and duplicate reads were marked with GATK (v4.2.3.0). Base quality score recalibration was performed using known variant sites from dbSNP and the Mills and 1000G gold standard indel resource [25]. Somatic variant calling was performed using Mutect2 (GATK v4.2.3.0) with matched normal controls, a panel of normals and the gnomAD germline resource. Variants with a population allele frequency greater than 1% were excluded. Variant calls were filtered using FilterMutectCalls. Additional filtering was performed using VariantFiltration with the following criteria: Quality by Depth (QD) < 2.0, Phred‐scaled variant quality score (QUAL) < 30, Fisher Strand bias (FS) > 200, strand odds ratio (SOR) > 10, or Read Position Rank Sum (ReadPosRankSum) < −20. Variant annotation was performed using SnpEff (v5.1). Variants annotated as HIGH or MODERATE impact were prioritised for review, whereas variants classified as LOW or MODIFIER were generally excluded from further analysis. Variants are reported based on the MANE Select transcript where available. Following exome‐wide variant calling and filtering, variants in IGLL5, MYD88, NOTCH2, CXCR4, CD79B, ARID1A, KLF2, CARD11, KMT2D and TNFAIP3 were specifically reviewed as an in silico gene panel based on their reported relevance to CAD and related B‐cell lymphoproliferative disorders [8, 26, 27]. Overall exome sequencing metrics, including total read depth, mean target coverage and target coverage at ≥ 20× depth, were assessed using CollectHsMetrics in GATK v4.2.3.0. Coverage of specific genes and hotspot regions was calculated from BAM files using samtools depth. Variants with comparable variant allele fractions in matched normal samples, marked strand bias, or insufficient sequencing support were excluded as likely germline variants or sequencing artefacts. All candidate variants were manually inspected using IGV to exclude potential sequencing and mapping artefacts. The whole‐exome sequencing data generated in this study have been deposited in the JGA (accession numbers: JGAS000791 and JGAS000869). Access to the data was controlled in accordance with ethical approval. Exploratory copy number analysis was performed using CNVkit (v0.9.13) based on whole‐exome sequencing read‐depth data and the Agilent SureSelect Human All Exon V6 target intervals, with sequencing reads aligned to the GRCh38 reference genome [28]. Results Clinical and Laboratory Features The clinical and laboratory characteristics of the eight patients are summarised in Table 1. The cohort comprised six men and two women, with a median age of 58 years (range, 49–80). All patients presented with cold‐induced symptoms, and none exhibited clinical or radiological evidence of an underlying lymphoproliferative disorder at diagnosis. Anaemia was present in all the patients, with a median haemoglobin concentration of 10.4 g/dL (range, 6.7–12.6). Lactate dehydrogenase levels were mildly to moderately elevated, with a median value of 234 U/L (range, 126–358). The direct antiglobulin test was positive for C3d in all patients. Cold agglutinin titres ranged from 1:64 to > 1:2048, with a median titre of 1:2048. A monoclonal IgM component was detected in six of eight patients (75%), predominantly IgM‐κ, whereas no monoclonal protein was identified in one patient, and data were unavailable for another. All patients reported symptoms triggered or exacerbated by cold exposure. TABLE 1Clinical and laboratory characteristics.No.AgeSexHb (g/dL)LDH (U/L)M proteinDAT (C3d)a CA titreUnderlying disease b Cold‐induced symptomsc 156Female10.6221IgM‐κPositive1:512NonePositive250Female12.6126NegativePositive1:2048NonePositive380Male11.5321IgM‐κPositive> 1:2048NonePositive456Male9.5358N/APositive1:128NonePositive574Male10.4254IgM‐κPositive> 1:2048NonePositive659Male8.6138IgM‐κPositive> 1:2048NonePositive760Male6.7246IgMPositive1:64NonePositive849Male10.4191IgM‐κPositive1:256NonePositiveAbbreviations: CA, cold agglutinin; DAT, direct antiglobulin test; Hb, haemoglobin; LDH, lactate dehydrogenase; N/A, not available. a DAT (C3d) indicates the results of the direct antiglobulin test specific for complement component C3d. b Underlying disease indicates conditions known to be associated with secondary cold agglutinin syndrome, including clinical or radiological findings that suggest overt lymphoma (such as lymphadenopathy or splenomegaly), autoimmune diseases, chronic infections, or other malignant disorders; none were identified at diagnosis. c Cold‐induced symptoms were defined as patient‐reported symptoms induced or exacerbated by cold exposure, including acrocyanosis of the extremities, nasal tip, and auricles; sensory disturbances; and Raynaud‐like phenomena. Clinical and laboratory characteristics. Abbreviations: CA, cold agglutinin; DAT, direct antiglobulin test; Hb, haemoglobin; LDH, lactate dehydrogenase; N/A, not available. DAT (C3d) indicates the results of the direct antiglobulin test specific for complement component C3d. Underlying disease indicates conditions known to be associated with secondary cold agglutinin syndrome, including clinical or radiological findings that suggest overt lymphoma (such as lymphadenopathy or splenomegaly), autoimmune diseases, chronic infections, or other malignant disorders; none were identified at diagnosis. Cold‐induced symptoms were defined as patient‐reported symptoms induced or exacerbated by cold exposure, including acrocyanosis of the extremities, nasal tip, and auricles; sensory disturbances; and Raynaud‐like phenomena. IGH/IGK Gene Analysis Immunoglobulin heavy and light chain gene rearrangements were analysed in all eight patients. Among the seven cases with productive IGH rearrangements, six of seven patients (85.7%) used IGHV4‐34 (Table 2), whereas one patient (Case 1) harboured an IGHV3‐33 rearrangement. The median germline homology to the closest reference sequence, as determined by IMGT, the international ImMunoGeneTics information system, was 94.7% (range, 93.2%–98.3%), and the median CDR3 length was 16 amino acids (range, 14–20). The coverage of FR1 was sufficient to evaluate the canonical Ala‐Val‐Tyr (AVY) motif in four of the six IGHV4‐34‐positive cases. The AVY motif was preserved in four cases (Cases 3, 4, 6 and 7). In two cases (Cases 2 and 5), the sequencing reads did not extend sufficiently upstream to allow a reliable assessment of the FR1 region. Somatic hypermutation affecting the N‐glycosylation motif within the CDR2 region was identified in one of six IGHV4‐34‐positive cases (16.7%). In the remaining cases, mutations were detected in residues flanking the motif. Mutations involving the Lys‐Leu‐Ser (KLS) sequence in FR3 were observed in three of six cases (50.0%). Analysis of the immunoglobulin light chain genes revealed IGKV3‐20 rearrangements in two patients (25%) and an IGKV1‐17 rearrangement in one patient (12.5%). No productive IGKV rearrangement was detected in the remaining cases. The germline homology of the IGKV rearrangements ranged from 97.3% to 100%. TABLE 2Characteristics of the rearranged immunoglobulin heavy and light chain genes.NoChainVDJHomology (%)CDR3 length (aa)CDR3 amino acid sequences1IGH IGHV3‐33 D3‐22 J6 94.320AKEGGDSXAWDGAYYYDXXV1IGK IGKV1‐17 — J4 97.39LQHNTFPLT2IGH IGHV4‐34 D2‐21 J6 93.815ARAYCGGGDCDYYYYG3IGH IGHV4‐34 D3‐22 J4 93.214ARGRGGYYDRSGFD4IGH IGHV4‐34 D2‐8, D3‐16 Not determined95.2Not determinedNot determined5IGH IGHV4‐34 D3‐10 J4 94.714ARPWGSGPGSYLNY5IGK IGKV3‐20 —Not determined1006QQYGSS6IGH IGHV4‐34 D6‐6 J4 98.317ARVGYSSSEEPHCPFDY7IGH IGHV4‐34 D2‐2 J6 97.618ARGVEYQLLEAYYYGMDV7IGK IGKV3‐20 —Not determined99.66QQYGXS Note: Homology (%) indicates nucleotide sequence identity to the closest germline gene, as determined by IMGT analysis. X indicates an undetermined amino acid.Abbreviations: aa, amino acids; CDR3, complementary determining region 3; D, diversity gene; IGH, immunoglobulin heavy chain; IGK, immunoglobulin kappa light chain; J, joining gene; V, variable gene. Characteristics of the rearranged immunoglobulin heavy and light chain genes. Note: Homology (%) indicates nucleotide sequence identity to the closest germline gene, as determined by IMGT analysis. X indicates an undetermined amino acid. Abbreviations: aa, amino acids; CDR3, complementary determining region 3; D, diversity gene; IGH, immunoglobulin heavy chain; IGK, immunoglobulin kappa light chain; J, joining gene; V, variable gene. Whole‐Exome Sequencing Analysis Whole‐exome sequencing was successfully performed in all eight patients. No pathogenic variants were detected in MYD88 or CXCR4 in any patient. Sequencing quality metrics were summarised in Tables S2–S4. The median sequencing depth at the MYD88 L265P hotspot was 316× (range, 183–729×), and no variant‐supporting reads were detected in any sample (Table S3). High‐confidence somatic variants were identified in four cases (Table 3). These included two truncating KMT2D mutations, one ARID1A frameshift mutation and one CARD11 missense mutation. In addition, clustered low‐variant allele frequency IGLL5 missense variants were observed in one case (Table S5). The two closely spaced IGLL5 variants were phased in cis based on read‐level inspection. WES‐based copy number analysis did not identify recurrent chromosomal gains involving Chromosomes 3, 12, or 18 (Table S6), although a possible gain of Chromosome 3 was observed in one case. Collectively, the mutational profile of this cohort was characterised by the absence of canonical MYD88 and CXCR4 mutations associated with LPL, despite the presence of additional mutations involving epigenetic regulators and NF‐κB‐related genes in a subset of cases. TABLE 3Somatic mutations identified by whole‐exome sequencing.CaseGeneTranscriptEffectProtein changeVAF (%)DepthAlt reads3 KMT2D NM_003482.3Stop‐gainp.Glu4434*5.9626375 ARID1A NM_006015.6Frameshiftp.Glu2023fs3.420775 KMT2D NM_003482.3Stop‐gainp.Gln1178*6.1261167 CARD11 NM_032415.6Missensep.Leu245Pro45.08036 Note: * indicates a stop codon.Abbreviation: VAF, variant allele frequency. Somatic mutations identified by whole‐exome sequencing. Note: * indicates a stop codon. Abbreviation: VAF, variant allele frequency. Bone Marrow Pathology Bone marrow pathological findings are summarised in Figure 1. Five of the eight patients (62.5%; Cases 2, 3, 4, 5 and 8) had pathological findings typical of CAD. A representative case of CAD is presented (Case 4, Figure 1A–D). In this case, scattered small nodular aggregates composed of small B cells were observed (Figure 1B). The small B cells had round nuclei, variably clumped chromatin and inconspicuous nucleoli. The plasma cells were sparse and were distributed mainly around, rather than within, the aggregates (Figure 1C). Plasma cells infiltrating the bone marrow were rare and accounted for less than 5% of the nucleated cells. No lymphoplasmacytoid cells were identified, and no increase in mast cells was observed within the aggregates. IgM expression in B cells and plasma cells was weak or heterogeneous (Figure 1D). Bone marrow pathological findings in typical cold agglutinin disease and lymphoplasmacytic lymphoma‐like cases. Case 4 was representative of typical CAD. In CAD, lymphocytes form small nodular aggregates within the bone marrow (A; HE, ×200). These aggregates consist predominantly of small B cells (B; CD20, ×200). The plasma cells are sparse and are distributed mainly at the periphery of the aggregates (C; CD138, ×200). In CAD, IgM expression is heterogeneous and is observed in a subset of plasma cells (D; IgM, ×200). Cases 6 and 7 were LPL‐like cases. In these cases, lymphocyte‐predominant nodular aggregates were observed (E, I; HE, ×200), and numerous aggregates of varying sizes were present throughout the bone marrow, which is consistent with bone marrow involvement by lymphoma. In the LPL‐like cases, the aggregates consisted predominantly of B cells (F, J; CD20, ×200); however, clusters of lymphoplasmacytoid cells and plasma cells were distributed at the periphery of the nodules (G, K; CD138, ×200). The distribution of plasma cells more closely resembled that seen in CAD. IgM expression was strongly positive in B cells, lymphoplasmacytoid cells and plasma cells (H, L; IgM, ×200). A representative case of typical nodular‐infiltrative LPL is shown. Nodular aggregates composed of small lymphocytes, lymphoplasmacytoid cells, and plasma cells were observed (M; HE, ×200). B cells proliferated throughout the aggregates (N; CD20, ×200). Lymphoplasmacytoid cells and plasma cells were distributed throughout the nodules, with prominent involvement of the central areas (O; CD138, ×200). B cells, lymphoplasmacytoid cells, and plasma cells within the nodules showed strong IgM expression (P; IgM, ×200). In two patients (Cases 6 and 7), multiple small and large nodular aggregates were observed throughout the bone marrow, occupying more than 50% of the bone marrow, which is consistent with bone marrow involvement by lymphoma (Figure 1E, I). In some areas, a diffuse infiltrative pattern was observed. Normal haematopoietic elements were markedly reduced. These aggregates consisted predominantly of small B cells (Figures 1F, J) and were accompanied by increased numbers of lymphoplasmacytoid cells and plasma cells at the periphery of the aggregates (Figure 1G, K). The overall features closely resembled those of CAD. More than 5% of the nucleated cells were plasma cells. Both cases showed plasmacytoid differentiation with Dutcher bodies and increased numbers of haemosiderin‐laden histiocytes. Mast cells within the aggregates were extremely scarce. IgM expression was strong in B cells, lymphoplasmacytoid cells and plasma cells in these patients (Figure 1H, L). These cases of disease were considered LPL‐like and were also thought to harbour features of CAD. The initial pathological diagnosis was LPL; however, no MYD88 L265P mutation was detected. No CD5‐positive or CD10‐positive clonal B‐cell population suggestive of a specific lymphoma subtype was identified by flow cytometry. For comparison, a representative case of LPL is shown in Figure 1M–P. In this case, multiple nodular aggregates composed of B cells were observed (Figure 1M, N). In contrast to the CAD and LPL‐like cases, lymphoplasmacytoid cells and plasma cells were not predominantly distributed at the periphery of the aggregates but were located mainly in the central areas of the aggregates (Figure 1O). In addition, increased infiltration of mast cells and haemosiderin‐laden histiocytes was observed within the aggregates. IgM expression was strongly positive in B cells, lymphoplasmacytoid cells and plasma cells (Figure 1P). Discussion In this study, we performed integrated pathological, immunogenetic and genomic analyses of eight patients with cold agglutinin–associated disorders to characterise diagnostically challenging cases. Most cases exhibited the established genetic hallmarks of CAD, including highly restricted IGHV4‐34 usage [1, 4, 7]. Notably, some cases fulfilled the morphological criteria for LPL but lacked its defining genetic features and instead exhibited CAD‐like immunogenetic and mutational profiles. These findings highlight a clinically relevant genotype–phenotype discordance and challenge the strict conceptual separation between CAD and CAS according to the fifth edition of the WHO classification. CAD is now recognised as a distinct bone marrow‐based clonal B‐cell lymphoproliferative disorder with relatively uniform histopathological features [4, 14]. Typical CAD is characterised by intraparenchymatous small nodular aggregates of CD20‐positive B lymphoid cells with limited plasmacytic differentiation. Plasma cells are usually sparse and distributed mainly around the lymphoid aggregates. IgM expression in lymphocytes and plasma cells is typically weak or heterogeneous [4, 14]. In the present study, the pathological findings in five of the eight patients were fully consistent with this established spectrum. In contrast, two patients had more extensive nodular infiltrates occupying more than 50% of the bone marrow, with morphological features resembling those of LPL. These included increased numbers of plasma cells and lymphoplasmacytoid cells peripheral to the aggregates, plasmacytoid differentiation with Dutcher bodies and strong IgM expression in lymphocytes and plasma cells. From a purely morphological standpoint, these features fulfil the diagnostic criteria for low‐grade B‐cell lymphoma and mandate classification as CAS under the current WHO framework. In the LPL control case, CD138‐positive plasma cells were predominantly located within the nodular aggregates, unlike the peripheral distribution observed in CAD and CAD‐like cases. In addition, mast cell proliferation, a common feature of LPL, was not observed in the CAD‐like cases. These findings suggest that the CAD‐like cases retain pathological features overlapping with CAD rather than true LPL [15, 29, 30, 31, 32]. The immunoglobulin gene repertoire provides further biological support for these observations [1, 13, 33]. In our cohort, IGHV4‐34 was identified in 85.7% of cases, consistent with previous reports and contrasting with the IGHV3 predominance typically observed in LPL. Somatic hypermutation patterns affecting I‐antigen binding motifs, including the AVY sequence in FR1 and the N‐glycosylation motif in CDR2, varied among patients but were not segregated between morphologically typical CAD and lymphoma‐like cases. These findings support the concept that these cases share a common immunogenetic background [1, 34]. Whole‐exome sequencing further reinforced the genetic signature of CAD in this study. No pathogenic variants were detected in MYD88 or CXCR4, the genetic hallmarks of LPL [6, 7]. Even in cases showing LPL‐like pathological features, canonical MYD88 and CXCR4 mutations were absent. These findings contradict a diagnosis of biologically true LPL. Moreover, none of the cases harboured mutations characteristic of marginal zone lymphoma, such as those in NOTCH2, KLF2, or TNFAIP3 [35, 36, 37]. These findings suggest that genetic profiling may better reflect disease identity than histomorphology alone. The IGLL5 mutations identified in the present cohort warrant cautious interpretation. Furthermore, no CD10‐positive clonal B‐cell population was identified by flow cytometry, arguing against an underlying follicular centre cell lymphoma. Moreover, the IGLL5 mutations observed in this study are best interpreted as passenger events, as IGLL5 is a well‐known off‐target of AID‐mediated somatic hypermutation in GC‐experienced B cells [9, 11, 12]. Their detection supports a GC‐experienced B‐cell origin, which is not typical of LPL. The disagreement between the pathological and genetic findings observed in this study raises important questions regarding the classification of cold agglutinin–associated disorders. Under the current WHO criteria, the presence of lymphoma‐like infiltrates necessitates a diagnosis of CAS, irrespective of the underlying genetic landscape. However, our observations suggest that clonal B‐cell proliferation in CAD may, in some cases, exhibit morphological features that satisfy the diagnostic thresholds for lymphoma without representing a biologically distinct lymphoid neoplasm. Whether such cases represent an expanded spectrum of CAD or a distinct biological entity remains uncertain. From a clinical perspective, this distinction is not merely semantic. Classification as CAD or CAS influences diagnosis and therapeutic considerations, particularly with the advent of complement‐directed therapies developed specifically for CAD. Although our study was not designed to assess treatment outcomes, it highlights the need for diagnostic algorithms that integrate molecular features with conventional pathology to better reflect underlying disease biology and to minimise the risk of misclassification. This study has several limitations, including its small sample size and retrospective design. In addition, pathological evaluation was limited to bone marrow aspirate clot sections, which limited comprehensive assessment of marrow architecture and strict application of WHO/ICC histopathological criteria. Nevertheless, the consistent immunogenetic and genomic patterns observed across morphologically heterogeneous cases support the biological relevance of our findings. In summary, we demonstrate that CAD‐like genetic and immunogenetic signatures can be present even in cases with lymphoma‐like morphological features that are currently classified as CAS. These results challenge the strict conceptual separation between CAD and CAS in the current WHO classification and support a diagnostic framework that incorporates genetic hallmarks alongside conventional pathology to more accurately reflect the biological spectrum of cold agglutinin–associated lymphoproliferative disorders. Further studies are warranted to refine disease definitions and to determine how molecular findings should inform future diagnostic classifications and clinical decision‐making. Author Contributions Y.S. performed the experiments, analysed the data and drafted the manuscript. D.K. designed the study, performed the experiments, analysed the data and wrote the manuscript. S.Y. and Y.O. analysed the pathological data. K.S. and M.F. performed the experiments. Y.H. and T.I. supervised the research. All the authors approved the final manuscript. Funding This work was supported by a Grant‐in‐Aid for Scientific Research (C) from the Japan Society for the Promotion of Science (D.K. and T.I.), a grant from the Takeda Science Foundation (D.K.) and funding from the Uehara Memorial Foundation (T.I.). Ethics Statement The study was performed with the approval of an institutional review board, and written informed consent was obtained from the patients. Conflicts of Interest Takayuki Ikezoe: Honoraria, Asahi Kasei. The other authors declare no conflicts of interest. Supporting information Supporting Information: jha270327‐sup‐0001‐TableS1.docx Supporting Information: jha270327‐sup‐0002‐TableS2.docx Supporting Information: jha270327‐sup‐0003‐TableS3docx Supporting Information: jha270327‐sup‐0004‐TableS4.docx Supporting Information: jha270327‐sup‐0005‐TableS5.docx Supporting Information: jha270327‐sup‐0006‐TableS6.docx Acknowledgements The authors are grateful for the contributions of all the participating physicians, staff, and patients. Data Availability Statement Whole‐exome sequencing analysis data were deposited in the Japanese Genotype–phenotype Archive database (accession numbers: JGAS000791 and JGAS000869). References Associated Data Supplementary Materials Supporting Information: jha270327‐sup‐0001‐TableS1.docx Supporting Information: jha270327‐sup‐0002‐TableS2.docx Supporting Information: jha270327‐sup‐0003‐TableS3docx Supporting Information: jha270327‐sup‐0004‐TableS4.docx Supporting Information: jha270327‐sup‐0005‐TableS5.docx Supporting Information: jha270327‐sup‐0006‐TableS6.docx Data Availability Statement Whole‐exome sequencing analysis data were deposited in the Japanese Genotype–phenotype Archive database (accession numbers: JGAS000791 and JGAS000869).