Clinicopathological features of adult T-cell leukemia/lymphoma with T-follicular helper phenotype.

AIMS: T-follicular helper (TFH) cells are effector T-cells that are crucial for B-cell selection and differentiation. T-cell lymphomas derived from TFH cells have distinct characteristics. Additionally, in the World Health Organization (WHO) classification 5th edition, three lymphomas were introduced as independent disease entities with TFH cell origin. We aimed to investigate the clinicopathological features of adult T-cell leukemia/lymphoma (ATLL) with a TFH phenotype (TFHP). METHODS AND RESULTS: We performed TFH immunohistochemistry analysis of five biomarkers for the biopsy specimen, with TFHP being indicated by a positive result for more than two markers. Among 75 cases of ATLL, 37.3% of them showed TFHP. Compared with cases of ATLL without TFHP, cases of ATLL with TFHP showed higher C-reactive protein levels (p = 0.0219) and increased high endothelial venule proliferation (p = 0.024). However, there were no significant between-group differences in overall survival as well as other clinical and morphological findings. Furthermore, there was no significant between-group difference in TFH markers and FOXP3 expression. CONCLUSION: Some patients with ATLL may present a TFHP, which should not preclude the diagnosis of ATLL. Although presenting a TFHP does not affect prognosis, it is important to identify cases of ATLL with a TFHP since it may inform future treatment strategies.

EBV+ nodal T/NK-cell lymphoma associated with clonal hematopoiesis and structural variations of the viral genome.

ABSTRACT: Epstein-Barr virus (EBV)-positive (EBV+) nodal T- and natural killer (NK)-cell lymphoma is a peripheral T-cell lymphoma (EBV+ nPTCL) that presents as a primary nodal disease with T-cell phenotype and EBV-harboring tumor cells. To date, the genetic aspect of EBV+ nPTCL has not been fully investigated. In this study, whole-exome and/or whole-genome sequencing was performed on 22 cases of EBV+ nPTCL. TET2 (68%) and DNMT3A (32%) were observed to be the most frequently mutated genes whose presence was associated with poor overall survival (P = .004). The RHOA p.Gly17Val mutation was identified in 2 patients who had TET2 and/or DNMT3A mutations. In 4 patients with TET2/DNMT3A alterations, blood cell-rich tissues (the bone marrow [BM] or spleen) were available as paired normal samples. Of 4 cases, 3 had at least 1 identical TET2/DNMT3A mutation in the BM or spleen. Additionally, the whole part of the EBV genome was sequenced and structural variations (SVs) were found frequent among the EBV genomes (63%). The most frequently identified type of SV was deletion. In 1 patient, 4 pieces of human chromosome 9, including programmed death-ligand 1 gene (PD-L1) were identified to be tandemly incorporated into the EBV genome. The 3' untranslated region of PD-L1 was truncated, causing a high-level of PD-L1 protein expression. Overall, the frequent TET2 and DNMT3A mutations in EBV+ nPTCL seem to be closely associated with clonal hematopoiesis and, together with the EBV genome deletions, may contribute to the pathogenesis of this intractable lymphoma.

Multimodal Gated Mixture of Experts Using Whole Slide Image and Flow Cytometry for Multiple Instance Learning Classification of Lymphoma.

In this study, we present a deep-learning-based multimodal classification method for lymphoma diagnosis in digital pathology, which utilizes a whole slide image (WSI) as the primary image data and flow cytometry (FCM) data as auxiliary information. In pathological diagnosis of malignant lymphoma, FCM serves as valuable auxiliary information during the diagnosis process, offering useful insights into predicting the major class (superclass) of subtypes. By incorporating both images and FCM data into the classification process, we can develop a method that mimics the diagnostic process of pathologists, enhancing the explainability. In order to incorporate the hierarchical structure between superclasses and their subclasses, the proposed method utilizes a network structure that effectively combines the mixture of experts (MoE) and multiple instance learning (MIL) techniques, where MIL is widely recognized for its effectiveness in handling WSIs in digital pathology. The MoE network in the proposed method consists of a gating network for superclass classification and multiple expert networks for (sub)class classification, specialized for each superclass. To evaluate the effectiveness of our method, we conducted experiments involving a six-class classification task using 600 lymphoma cases. The proposed method achieved a classification accuracy of 72.3%, surpassing the 69.5% obtained through the straightforward combination of FCM and images, as well as the 70.2% achieved by the method using only images. Moreover, the combination of multiple weights in the MoE and MIL allows for the visualization of specific cellular and tumor regions, resulting in a highly explanatory model that cannot be attained with conventional methods. It is anticipated that by targeting a larger number of classes and increasing the number of expert networks, the proposed method could be effectively applied to the real problem of lymphoma diagnosis.

Clinicopathological comparison between PTCL-TBX21 and PTCL-GATA3 in Japanese patients.

AIM: Peripheral T-cell lymphoma not otherwise specified (PTCL-NOS) is a heterogeneous disease that can be classified into the PTCL-TBX21 and PTCL-GATA3 subtypes. METHODS: In this study, we compared the clinicopathological features of PTCL-NOS in a Japanese cohort, classified using an IHC algorithm. RESULTS: One hundred patients with PTCL-NOS were categorized as having PTCL-TBX21 (n = 55), PTCL-GATA3 (n = 24), or PTCL-unclassified (n = 21). When comparing PTCL-TBX21 and PTCL-GATA3, PTCL-TBX21 showed significantly lower CD4 positivity (p = 0.047), lower counts of high endothelial venules (p = 0.032), and a tendency for a better response to initial treatment (p = 0.088). Gene expression analysis using the nCounter system showed higher expression of tumor immunity-related genes, such as PD-L1, LAG3, and IDO1, in PTCL-TBX21 than in PTCL-GATA3. PTCL-GATA3 had significantly worse overall survival (OS) than those with PTCL-TBX21 (p = 0.047), although a similar tendency was observed for progression-free survival (PFS) (p = 0.064). PTCL-GATA3 was a prognostic factor for OS in univariate analysis (HR 2.02; 95% CI, 1.09-3.77; p = 0.027), although multivariate analysis did not show significance (HR 2.07; 95% CI, 0.93-4.61; p = 0.074). In the PFS analysis, PTCL-GATA3 was an independent prognostic factor by univariate analysis (HR 1.96; 95% CI, 1.08-3.56; p = 0.027) and multivariate analysis (HR 2.34; 95% CI, 1.07-5.11; p = 0.032). CONCLUSION: The classification of PTCL-NOS into PTCL-TBX21 and PTCL-GATA3 is useful for predicting the prognosis of Japanese patients and stratifying the administration of tumor immune checkpoint inhibitors in clinical practice.

Primary renal peripheral T-cell lymphoma, not otherwise specified, treated with partial nephrectomy.

Introduction: Renal involvement by non-Hodgkin's lymphoma is very rare, and the kidney as the primary site of this lymphoma is much more uncommon. We report a case of primary renal peripheral T-cell lymphoma, not otherwise specified, treated with partial nephrectomy. Case presentation: A 63-year-old man was hospitalized with coronavirus infectious disease, emerged in 2019 in the emergency department. Computed tomography examination showed a 2-cm renal mass in the right kidney. Abdominal enhanced computed tomography examination revealed that the noted mass showed good enhancement in the corticomedullary phase and washout in the nephrogenic phase. No metastatic lesions were found. He was diagnosed as having cT1aN0M0 renal cell carcinoma, and robotic-assisted partial nephrectomy was carried out. The pathological diagnosis was peripheral T-cell lymphoma, not otherwise specified. He has been followed for 20 months after robotic-assisted partial nephrectomy without additional treatment and recurrence. Conclusion: We experienced a primary renal peripheral T-cell lymphoma, not otherwise specified that was followed up without treatment after surgery.

18F-FDG PET/CT imaging of IgG4-producing MALT lymphoma with multiple site involvement.

18F-FDG PET/CT is regarded as a modality utilized for the purpose of lesion localization, staging and assessment of treatment response in patients with lymphoma. However, it is difficult that we diagnose among multifocal lymphoma, IgG4-related disease (IgG4-RD), or a combination of both conditions when confronted with multiple sites of 18F-FDG uptake with heightened serum IgG4 levels. We present a case of a 72-year-old male who was suspected of Sjögren's syndrome based on symptoms of xerostomia accompanied by swelling of the bilateral upper eyelid and salivary glands. Following a diagnostic biopsy that revealed mucosa-associated lymphoid tissue (MALT) lymphoma as a possible finding, 18F-FDG PET/CT was conducted, which demonstrated multiple sites of 18F-FDG accumulation. While multifocal MALT lymphoma was initially suspected, the coexistence of IgG4-RD could not be definitively ruled out due to the elevated serum IgG4 levels. Subsequent histopathological and immunohistochemical examinations confirmed the diagnosis of IgG4-producing MALT lymphoma. After receiving systemic therapy with rituximab, the swelling of the bilateral upper eyelid and parotid glands resolved upon visual examination, and the serum IgG4 levels returned to within the normal range in a few months. No new lesions were detected during the subsequent follow-up examinations conducted over a period of 3 years.

Co-expression of regulatory B-cell markers, transforming growth factor ß and interleukin-10 as a prognostic factor in diffuse large B-cell lymphoma.

Regulatory B cells (Bregs) suppress antitumor immunity by producing anti-inflammatory cytokines such as transforming growth factor ß (TGF-ß) and interleukin-10 (IL-10) and promoting tumor growth. It is unknown whether diffuse large B-cell lymphoma (DLBCL), a common subtype of B-cell malignancy, exhibits characteristics similar to those of Bregs. This study aimed to clarify the features of DLBCLs carrying Breg markers. In 123 DLBCL cases, we evaluated TGF-ß and IL-10 expression in tumor biopsy samples using immunohistochemical staining and retrospectively analyzed their clinicopathological characteristics. Fifteen cases (12.2 %) classified as Breg-type DLBCL were positive for both TGF-ß and IL-10. Breg-type DLBCL is mainly classified as having activated B cell-like cells of origin. Breg-type DLBCL cases showed significantly worse progression-free survival and overall survival (OS) than other DLBCL cases (P = 0.0016 and P = 0.042, respectively). In multivariate analysis, Breg-type DLBCL significantly affected OS (hazard ratio, 3.13; 95 % confidence interval 1.15-8.55; P = 0.025). Gene expression analysis showed that the expression of follicular dendritic cell-associated genes (FCER2, PIK3CD, FOXO1) was downregulated in Breg-type DLBCLs compared to other DLBCLs. These results suggest that the double expression of Breg markers, TGF-ß and IL-10, in tumor cells indicates a poor prognosis in DLBCL patients. Further studies evaluating genomic abnormalities could confirm the characteristics of Breg-type DLBCL.

The Detection of Neoplastic Cells Using Objective Cytomorphologic Parameters in Malignant Lymphoma.

Pathologic evaluation is the most crucial method for diagnosing malignant lymphomas. However, there are no established diagnostic criteria for evaluating pathologic morphology. We manually circled cell nuclei in the lesions of 10 patients with diffuse large B-cell lymphoma (DLBCL), follicular lymphoma, and reactive lymphadenitis. Seventeen parameters related to nuclear shape, color, and other characteristics were measured. We attempted to compare the statistical differences between these subtypes and extract distinctive disease-specific populations on the basis of these parameters. Statistically significant differences were observed between the different types of lymphoma for many of the 17 parameters. Through t-distributed stochastic neighbor embedding analysis, we extracted a cluster of cells that showed distinctive features of DLBCL and were not found in follicular lymphoma or reactive lymphadenitis. We created a decision tree to identify the characteristics of the cells within that cluster. Based on a 5-fold cross-validation study, the average sensitivity, specificity, and accuracy obtained were 84.1%, 98.4%, and 97.3%, respectively. A similar result was achieved using a validation experiment. Important parameters that indicate the features of DLBCL include Area, ConcaveCount, MaxGray, and ModeGray. By quantifying pathologic morphology, it was possible to objectively represent the cell morphology specific to each lymphoma subtype using quantitative indicators. The quantified morphologic information has the potential to serve as a reproducible and flexible diagnostic tool.

Increased expression of leucine-rich α-2 glycoprotein 1 as a predictive biomarker of favorable progression-free survival in meningioma.

Most meningiomas, which are frequent central nervous system tumors, are classified as World Health Organization (WHO) grade 1 because of their slow-growing nature. However, the recurrence rate varies and is difficult to predict using conventional histopathological diagnoses. Leucine-rich α-2 glycoprotein 1 (LRG1) is involved in cell signal transduction, cell adhesion, and DNA repair and is a predictive biomarker in different malignant tumors; however, such a relationship has not been reported in meningiomas. We examined tissue microarrays of histological samples from 117 patients with grade 1 and 2 meningiomas and assessed their clinical and pathological features, including expression of LRG1 protein. LRG1-high meningiomas showed an increased number of vessels with CD3-positive cell infiltration (P = 0.0328) as well as higher CD105-positive vessels (P = 0.0084), as compared to LRG1-low cases. They also demonstrated better progression-free survival (hazard ratio [HR] 0.11, 95% confidence interval [CI] 0.016-0.841) compared to LRG1-low patients (P = 0.033). Moreover, multivariate analysis indicated that high LRG1 expression was an independent prognostic factor (HR, 0.13; 95% CI, 0.018-0.991; P = 0.049). LRG1 immunohistochemistry may be a convenient tool for estimating the prognosis of meningiomas in routine practice. Further studies are required to elucidate the key role of LRG1 in meningioma progression.

Clinicopathological analysis of CD47 and signal regulatory protein alpha expression in myeloid sarcoma patients: CD47 expression is a favourable prognostic factor.

Myeloid sarcoma is a rare extramedullary haematopoietic malignancy. Interaction between CD47 and signal regulatory protein α (SIRPα) inhibits phagocytosis. CD47-positive tumours confer poor prognoses in various malignant tumours, including acute myeloid leukaemia. This study aimed to investigate the clinicopathological effects of CD47 and SIRPα expression in myeloid sarcoma. Immunohistochemistry (IHC) of CD47 and SIRPα was performed in 84 biopsy samples obtained from patients with myeloid sarcoma, some of which were CD47-positive. Patients were categorised into the following two groups based on IHC of SIRPα: those with SIRPα-positive neoplastic cells (nSIRPα) and, SIRPα expression on non-neoplastic stromal cells in tumour microenvironment (miSIRPα). In addition, patients with CD47 positivity had higher lymphocytic infiltration into the tumour microenvironment. Overall, these patients had significantly higher overall survival, however, no significant difference was observed in progression-free survival. No significant prognostic differences were observed between the nSIRPα and miSIRPα groups. This is the first study to demonstrate an association between CD47 expression and improved prognosis in myeloid sarcoma. Nonetheless, it will be necessary to conduct additional research on gene expression and genomic abnormalities to elucidate the corresponding pathogenesis of myeloid sarcoma.

The first autopsy case of Epstein-Barr virus-positive marginal zone lymphoma that deteriorated after COVID-19 vaccination.

This is the first autopsy case of Epstein-Barr virus-positive marginal zone lymphoma (EBV + MZL) with an other iatrogenic immunodeficiency-associated lymphoproliferative disorders (LPD) (methotrexate [MTX]-associated LPD) that deteriorated after the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccine. This case had a unique immunophenotype. A 71-year-old female patient with rheumatoid arthritis receiving MTX presented with fatigue 1 week after the SARS-CoV-2 vaccination. She was hospitalized due to hepatorenal dysfunction and pancytopenia. Computed tomography revealed systemic lymphadenopathy. Her physical condition deteriorated, and the patient died. The autopsy revealed systemic lymphadenopathy comprising medium-sized atypical lymphocytes and scattered Hodgkin/Reed-Sternberg (H/RS)-like cells. An immunohistochemical examination showed that atypical lymphocytes were positive for CD79a and MUM-1 and some were positive for CD20 and IRTA-1. H/RS-like cells were immunoreactive for CD30 and CD15 and ringed by T cells. Both cell types were positive for EBV-encoded small RNA. The majority of H/RS-like cells were positive for CD20, whereas a small number of CD3-positive cells were admixed. We herein presented the first autopsy case of EBV + MZL that deteriorated after the SARS-CoV-2 vaccination.

Angioimmunoblastic T-Cell Lymphoma after Treatment of Classic Hodgkin Lymphoma: A Case Report.

We report a case of a 24-year-old man who developed angioimmunoblastic T-cell lymphoma (AITL) after treatment for refractory lymphocyte-rich classic Hodgkin lymphoma (LR-CHL). This patient was treated with the BV+AVD (brentuximab vedotin, doxorubicin, vinblastine, and dacarbazine) protocol for LR-CHL but progressed before completing chemotherapy. The pathological imaging showed the typical findings of LR-CHL at the first onset and first progression. Rescue chemotherapy and high-dose chemotherapy combined with autologous hematopoietic stem cell transplantation (AHSCT) were performed for refractory LR-CHL, and complete remission was achieved. However, the recurrence was suspected 6 months after AHSCT. The pathological findings of the lymph node biopsy at this time were different from those of the previous two lymph node biopsies, demonstrating findings of AITL. The finding of the immunohistochemical staining and polymerase chain reaction results supported the diagnosis. Although it has been reported that the risk for the development of non-Hodgkin lymphoma after treatment for Hodgkin lymphoma is increased, most are B-cell lymphomas, and few cases of AITL have been reported. AITL is a type of peripheral T-cell lymphoma that generally occurs in middle-aged and elderly people and that rarely occurs in young people. Here, we were able to make an accurate diagnosis by performing re-examination even when recurrence of LR-CHL was suspected. As there are no detailed case reports of AITL developing into secondary non-Hodgkin lymphoma, here we report on an identified case.

Intratumoral thrombosis as a histological biomarker for predicting epidermal growth factor receptor alteration and poor prognosis in patients with glioblastomas.

PURPOSE: Intratumoral thrombosis is a specific finding in glioblastomas and considered the origin of palisading necrosis. Its distribution and contribution to the glioblastoma pathophysiology and systemic thrombosis are obscure, although deep vein thrombosis is a common complication in glioblastoma cases. METHODS: Clinicopathological and genetic analyses were performed on 97 glioblastoma tissue specimens to elucidate the role of thrombotic events and associated molecular abnormalities. RESULTS: Morphologically, intratumoral thrombosis was observed more frequently in vessels composed of single-layered CD34-positive endothelium and/or αSMA-positive pericytes in the tumor periphery, compared to microvascular proliferation with multi-channeled and pericyte-proliferating vessels in the tumor center. Intratumoral thrombosis was significantly correlated with the female sex, high preoperative D-dimer levels, and epidermal growth factor receptor (EGFR) amplification. The presence of one or more thrombi in 20 high-power fields was a predictive marker of EGFR amplification, with a sensitivity of 81.5% and specificity of 52.6%. RNA sequencing demonstrated that the group with many thrombi had higher EGFR gene expression levels than the group with few thrombi. The tumor cells invading along the vessels in the tumor periphery were positive for wild-type EGFR but negative for EGFRvIII, whereas the cells around the microvascular proliferation (MVP) in the tumor center were positive for both wild-type EGFR and EGFRvIII. Intratumoral thrombosis is an independent poor prognostic factor. CONCLUSIONS: Aberrant but exquisitely regulated EGFR can induce thrombosis in non-MVP vessels in the tumor invasion area and then promote palisading necrosis, followed by hypoxia, abnormal angiogenesis, and further tumor cell invasion.

Digital spatial profiling of CD4+ T cells in classic Hodgkin lymphoma.

Classic Hodgkin lymphoma (CHL) harbors a small number of Hodgkin-Reed-Sternberg (HRS) cells scattered among numerous lymphocytes. HRS cells are surrounded by distinct CD4+ T cells in a rosette-like manner. These CD4+ T cell rosettes play an important role in the tumor microenvironment (TME) of CHL. To elucidate the interaction between HRS cells and CD4+ T cell rosettes, we completed digital spatial profiling to compare the gene expression profiles of CD4+ T cell rosettes and other CD4+ T cells separated from the HRS cells. Immune checkpoint molecules including OX40, programed cell death-1 (PD-1), and cytotoxic T lymphocyte associated protein 4 (CTLA-4) expression was higher in CD4+ T cell rosettes compared to other CD4+ T cells. Immunohistochemistry confirmed variable PD-1, CTLA-4, and OX40 expression in the CD4+ T cell rosettes. This study introduced a new pathological approach to study the CHL TME, and provided deeper insight into CD4+ T cells in CHL.

HTLV-1 bZIP Factor-Induced Reprogramming of Lactate Metabolism and Epigenetic Status Promote Leukemic Cell Expansion.

Acceleration of glycolysis is a common trait of cancer. A key metabolite, lactate, is typically secreted from cancer cells because its accumulation is toxic. Here, we report that a viral oncogene, HTLV-1 bZIP factor (HBZ), bimodally upregulates TAp73 to promote lactate excretion from adult T-cell leukemia-lymphoma (ATL) cells. HBZ protein binds to EZH2 and reduces its occupancy of the TAp73 promoter. Meanwhile, HBZ RNA activates TAp73 transcription via the BATF3-IRF4 machinery. TAp73 upregulates the lactate transporters MCT1 and MCT4. Inactivation of TAp73 leads to intracellular accumulation of lactate, inducing cell death in ATL cells. Furthermore, TAp73 knockout diminishes the development of inflammation in HBZ-transgenic mice. An MCT1/4 inhibitor, syrosingopine, decreases the growth of ATL cells in vitro and in vivo. MCT1/4 expression is positively correlated with TAp73 in many cancers, and MCT1/4 upregulation is associated with dismal prognosis. Activation of the TAp73-MCT1/4 pathway could be a common mechanism contributing to cancer metabolism. SIGNIFICANCE: An antisense gene encoded in HTLV-1, HBZ, reprograms lactate metabolism and epigenetic modification by inducing TAp73 in virus-positive leukemic cells. A positive correlation between TAp73 and its target genes is also observed in many other cancer cells, suggesting that this is a common mechanism for cellular oncogenesis. This article is featured in Selected Articles from This Issue, p. 337.

[Methotrexate-related lymphoproliferative disease showing different histological findings at recurrence].

A 55-year old female patient was treated with methotrexate (MTX) and infliximab (IFX) for rheumatoid arthritis (RA). She experienced unknown fever, generalized lymphadenopathy, and liver tumors. Histological examination of the inguinal lymph node and a liver tumor resulted in the pathological diagnosis of classic Hodgkin lymphoma, with many Reed-Sternberg cells with the positivity of Epstein-Barr virus (EBV). She was diagnosed with MTX-related lymphoproliferative disorders (MTX-LPDs). She received chemotherapy after the cessation of MTX and IFX and achieved complete remission. RA showed recurrence after a while, and she was treated with steroids or other drugs. Six years after the chemotherapy, she experienced low-grade fever and anorexia. Whole computed tomography images showed an appendix tumor and enlargement of the surrounding lymph nodes. Appendectomy with the radical lymph nodes dissection was performed. The pathological diagnosis was diffuse large B-cell lymphoma, resulting in the clinical diagnosis of the relapse of MTX-LPD. EBV was negative at this point. The pathological findings of MTX-LPD may change at relapse; thus, biopsy should be considered when the relapse of MTX-LPD is suggested.

Integrated genetic and clinical prognostic factors for aggressive adult T-cell leukemia/lymphoma.

The prognosis of aggressive adult T-cell leukemia/lymphoma (ATL) is poor, and allogeneic hematopoietic stem cell transplantation (allo-HSCT) is a curative treatment. In order to identify favorable prognostic patients after intensive chemotherapy, and who therefore might not require upfront allo-HSCT, we aimed to improve risk stratification of aggressive ATL patients aged <70 years. The clinical risk factors and genetic mutations were incorporated into risk modeling for overall survival (OS). We generated the m7-ATLPI, a clinicogenetic risk model for OS, that included the ATL prognostic index (PI) (ATL-PI) risk category, and non-silent mutations in seven genes, namely TP53, IRF4, RHOA, PRKCB, CARD11, CCR7, and GATA3. In the training cohort of 99 patients, the m7-ATLPI identified a low-, intermediate-, and highrisk group with 2-year OS of 100%, 43%, and 19%, respectively (hazard ratio [HR] =5.46; P<0.0001). The m7-ATLPI achieved superior risk stratification compared to the current ATL-PI (C-index 0.92 vs. 0.85, respectively). In the validation cohort of 84 patients, the m7-ATLPI defined low-, intermediate-, and high-risk groups with a 2-year OS of 81%, 30%, and 0%, respectively (HR=2.33; P=0.0094), and the model again outperformed the ATL-PI (C-index 0.72 vs. 0.70, respectively). The simplified m7-ATLPI, which is easier to use in clinical practice, achieved superior risk stratification compared to the ATLPI, as did the original m7-ATLPI; the simplified version was calculated by summing the following: high-risk ATL-PI category (+10), low-risk ATL-PI category (-4), and non-silent mutations in TP53 (+4), IRF4 (+3), RHOA (+1), PRKCB (+1), CARD11 (+0.5), CCR7 (-2), and GATA3 (-3).

Transformer-based personalized attention mechanism for medical images with clinical records.

In medical image diagnosis, identifying the attention region, i.e., the region of interest for which the diagnosis is made, is an important task. Various methods have been developed to automatically identify target regions from given medical images. However, in actual medical practice, the diagnosis is made based on both the images and various clinical records. Consequently, pathologists examine medical images with prior knowledge of the patients and the attention regions may change depending on the clinical records. In this study, we propose a method, called the Personalized Attention Mechanism (PersAM) method, by which the attention regions in medical images according to the clinical records. The primary idea underlying the PersAM method is the encoding of the relationships between medical images and clinical records using a variant of the Transformer architecture. To demonstrate the effectiveness of the PersAM method, we applied it to a large-scale digital pathology problem involving identifying the subtypes of 842 malignant lymphoma patients based on their gigapixel whole-slide images and clinical records.

Case-based similar image retrieval for weakly annotated large histopathological images of malignant lymphoma using deep metric learning.

In the present study, we propose a novel case-based similar image retrieval (SIR) method for hematoxylin and eosin (H&E) stained histopathological images of malignant lymphoma. When a whole slide image (WSI) is used as an input query, it is desirable to be able to retrieve similar cases by focusing on image patches in pathologically important regions such as tumor cells. To address this problem, we employ attention-based multiple instance learning, which enables us to focus on tumor-specific regions when the similarity between cases is computed. Moreover, we employ contrastive distance metric learning to incorporate immunohistochemical (IHC) staining patterns as useful supervised information for defining appropriate similarity between heterogeneous malignant lymphoma cases. In the experiment with 249 malignant lymphoma patients, we confirmed that the proposed method exhibited higher evaluation measures than the baseline case-based SIR methods. Furthermore, the subjective evaluation by pathologists revealed that our similarity measure using IHC staining patterns is appropriate for representing the similarity of H&E stained tissue images for malignant lymphoma.