Chiu, April

Background: Flow cytometry (FC) is widely utilized for the identification of minimal residual disease (MRD) in B acute lymphoblastic leukemia (BALL), providing an essential laboratory indicator for prognostic assessment and therapeutic management. Unfortunately, the high level of expertise required, long manual analysis time, and complexity of FC data files limit the offering of this testing to selected reference laboratories. Methods: We developed an artificial intelligence (AI) pipeline to automatically enhance unaltered FC files, minimizing the expertise and manual analysis time needed to detect BALL-MRD on a single-tube 10-color panel (CD10, CD19, CD20, CD22, CD24, CD34, CD38, CD45, CD58, CD66c). Raw FC files corresponding to peripheral blood or bone marrow specimens from 171 BALL MRD positive cases (MRD < 5%, median = 0.11%, 107 patients), and 89 MRD-negative cases (63 patients) were processed through our CCADDAS (Clustering and Classification of All events, Dimensionality reduction, Downsampling and Aberrancy Scaling) pipeline on a cloud environment (Google Vertex AI). Automated processing steps included elimination of acquisition errors (FlowCut), state-of-the-art clustering (PARC), dimensionality reduction (UMAP), cluster-based anomaly detection compared to negative controls (24 bone marrow aspirates and 7 peripheral bloods), and cluster-informed downsampling with preservation of low-event subsets. In addition, a deep neural network (DNN) trained on expert-defined subpopulations from negative controls was included for automatic gating of normal subsets (Tensorflow v2). AI-enhanced data files were analyzed using a general purpose flow cytometry analysis software (Kaluza v 3.5, Beckman Coulter), and %MRD estimates compared to reported results based on expert analysis of original FC files. Results: Cluster-informed downsampling reduced the number of cells per case to be manually analyzed from 1.2 million to 155,884 cells on average (87% cellularity reduction); resulting in a smaller FC data file (from 64.5 MB to 13.7 MB; 79% data reduction). Importantly, low-level MRD events were adequately preserved after downsampling (median 100% retention for both MRD<0.01% and MRD 0.01-1%). True number of events were accurately estimated on gated subsets using an “upsampling factor” parameter [observed # events x mean (upsampling factor)]. Gating of normal subsets was completely automated using a DNN classifier parameter. Identification of MRD was aided by an AI-generated “aberrancy scale” parameter that discriminated BALL (median: 991) from both benign mononuclear cells (median: 486, p < 0.0001) and background B-lymphoid cells (median: 379, p < 0.0001), with a performance (AUC: 0.98 and 0.94 for mononuclear and B-lymphoid cells, respectively) superior to CD20 (0.54 and 0.69), CD38 (0.60 and 0.81), CD45 (0.75 and 0.74), CD58 (0.52 and 0.89), CD66c (0.56 and 0.72) and all other antigens studied. The use of AI-enhanced files reduced manual analysis time from 14.16 minutes (SD = 3.7) to 1.05 minutes (SD = 0.6) per case, on average (p < 0.0001) (93% reduction). BALL MRD was detected in all positive cases above clinically relevant threshold (≥0.01%, as recommended by NCCN guidelines), and in 28 of 32 (88%) cases below this threshold, with excellent quantitative correlation with conventional analysis on original FC files (linear regression slope = 1.18, intercept = 0.04%, R squared = 0.92, P < 0.0001). Conclusion: We introduce a largely unsupervised AI pipeline that transforms raw BALL-MRD FC data into a markedly smaller, AI-annotated and software-agnostic FC file, including comparison to normal controls. This CCADDAS pipeline simplifies and accelerates detection of BALL-MRD in clinical diagnostics, reducing number of cells analyzed by 87% and manual analysis time by 93%, without impacting test performance. Moreover, CCADDAS can be utilized for any labroatory-developed BALL-MRD assay, and its small-sized export is compatible with any clinical FC software, computing platform and analysis strategy. Adoption of CCADDAS is likely to facilitate the implementation of BALL-MRD FC analysis by more clinical laboratories.

Background: Flow cytometry (FC) is widely utilized for the identification of minimal residual disease (MRD) in chronic lymphocytic leukemia (CLL), providing an essential laboratory indicator for prognostic assessment and therapeutic management. Unfortunately, the high level of expertise required, long manual analysis time, and complexity of FC data files limit the offering of this testing to selected reference laboratories. Methods: We developed an artificial intelligence (AI) pipeline to automatically enhance unaltered FC files, minimizing the expertise and manual analysis time needed to detect CLL MRD on a single-tube 10-color panel (CD5, CD19, CD20, CD22, CD38, CD43, CD45, CD200, kappa and lambda). Raw FC files corresponding to peripheral blood or bone marrow specimens from 166 CLL MRD positive cases (MRD < 5%, median = 0.17%, 122 patients), and 61 MRD-negative cases (45 patients) were processed through our CCADDAS (Clustering and Classification of All events, Dimensionality reduction, Downsampling and Aberrancy Scaling) pipeline on a cloud environment (Google Vertex AI). Automated processing steps included elimination of acquisition errors (FlowCut), state-of-the-art clustering (PARC), dimensionality reduction (UMAP), cluster-based anomaly detection compared to negative controls (15 bone marrow aspirates and 14 peripheral bloods), and cluster-informed downsampling with preservation of low-event subsets. In addition, a deep neural network (DNN) trained on expert-defined subpopulations from negative controls was included for automatic gating of normal subsets (Tensorflow v2). AI-enhanced data files were analyzed using a general purpose flow cytometry analysis software (Kaluza v 3.5, Beckman Coulter), and %MRD estimates compared to reported results based on expert analysis of original FC files. Results: Cluster-informed downsampling reduced the number of cells per case to be manually analyzed from 1.1 million to 165,010 cells on average (85% cellularity reduction); resulting in a smaller FC data file (from 62.1 MB to 15.2 MB; 75% data reduction). Importantly, low-level MRD events were adequately preserved after downsampling (median 100% retention for both MRD<0.01% and MRD 0.01-1%). True number of events were accurately estimated on gated subsets using an “upsampling factor” parameter [observed # events x mean (upsampling factor)]. Gating of normal subsets was completely automated using a DNN classifier parameter. Rapid identification of suspected MRD was aided by an AI-generated “aberrancy scale” parameter that discriminated CLL MRD (mean ± 2SD: 969 ± 144) from background B-lymphoid cells (mean ± 2SD: 395 ± 401) (p<0.0001) (AUC = 0.98), with a performance superior to CD5 (AUC = 0.93), CD20 (AUC = 0.83), CD43 (AUC = 0.8) and CD200 (AUC = 0.7). Immunoglobulin light chain-positive CLL MRD was visually identified on a pre-calculated UMAP plot as a single-colored cluster, after dual coloring for kappa and lambda. The use of AI-enhanced files reduced manual analysis time from 11.6 minutes (SD = 4.5) to 0.8 minutes (SD = 0.4) per case, on average (p < 0.01) (93% reduction). CLL MRD was detected in all positive cases above clinically relevant threshold (≥0.01%, as recommended by NCCN guidelines), and in 17 of 21 (81%) cases below this threshold, with excellent quantitative correlation with conventional analysis on original FC files (linear regression slope = 1.02, intercept = 0.001%, R squared = 0.98, P < 0.0001). Conclusion: We introduce a largely unsupervised AI pipeline that transforms raw CLL-MRD FC data into a markedly smaller, AI-annotated and software-agnostic FC file, including comparison to normal controls. This CCADDAS pipeline simplifies and accelerates detection of CLL-MRD in clinical diagnostics, reducing number of cells analyzed by 85% and manual analysis time by 93%, without impacting test performance. Moreover, CCADDAS can be utilized for any laboratory-developed CLL-MRD assay, and its small-sized export is compatible with any clinical FC software, computing platform and analysis strategy. Adoption of CCADDAS is likely to facilitate the implementation of CLL MRD FC analysis by more clinical laboratories.

Classic Hodgkin Lymphoma (cHL) is characterized by the presence of abnormal large cells known as Hodgkin Reed-Sternberg (HRS) cells. Curative intent chemotherapy has a 90% response rate, but for refractory patients, current immunotherapies remain insufficient. Ongoing research aims to enhance therapy and patient outcomes by focusing on the tumor microenvironment (TME), a complex environment that includes diverse non-malignant cells, including immune cells, which constitute a critical component of the TME. Utilizing multiplex analysis to comprehend these intricate interactions can pave the way for identifying novel strategies to combat cancer more effectively. CD30-targeted therapies and PD1/PDL1 inhibitors are under investigation clinically and could potentially benefit from adding checkpoint markers such as TIM-3 to enhance patient outcomes by addressing T-cell exhaustion. Such combination strategies will require assays and analysis strategies that can describe both spatial and multi-marker data. Multiplex immunofluorescence (mIF) is a valuable tool for analyzing spatial biology, patient selection, mechanism of action, pharmacodynamics and biomarker assays predictive of response. Describing the relationship between T cell exhaustion and response in cHL may guide the development of targeted treatments aimed at restoring the immune system's ability to recognize and eliminate cancer cells. Twenty-five to thirty percent of cHL patients with advanced stages do not respond to standard therapies. To understand the mechanisms of resistance, we developed a mIF panel (CD4, CD8, CD30, PD-1, PD-L1, and TIM3) for formalin-fixed and paraffin-embedded cHL tissues on the OPAL Tyramide Signal Amplification system and the Leica Bond RX. A set of commercially sourced cHL whole tissue samples (n=5) were used to develop and validate the panel, with qualitative evaluations by a Hematopathologist affirming that the mIF and 3, 3'-diaminobenzidine (DAB) staining yielded comparable results. Subsequently, a cohort of cHL samples (n=34) were stained and underwent exploratory image (Halo) and data (Microsoft Excel and R Studio) analyses. The image analysis strategy includes cell segmentation and phenotyping for all markers in the panel, identifying individual cell populations and multi-marker co-expressing cell populations. Additionally, spatial analysis describes the relationships between key phenotypes in the TME. The quantity of exhausted T cells, their average distance to HRS cells and the average count of exhausted T cells within different proximities of HRS cells will be presented. The development of this mIF assay for cHL demonstrates its consistent application at a mid-scale level across a diverse patient population with varying medical backgrounds and treatment histories. This approach provides a deeper understanding of immune response diversity and cellular composition within cHL. Future deployment of the panel against clinical trial samples with outcomes and responses will yield further insights.

Abstract Introduction: MYC+BCL2+ double expression (DE) is a biomarker for poor prognosis in diffuse large B-cell lymphoma (DLBCL), likely due to the cooperation between MYC and BCL2 oncoproteins in tumor cells. However, as generally MYC and BCL2 are separately assessed on two different slides by immunohistochemistry (IHC), MYC+ and BCL2+ results are from different cells. With the WHO-recommended cutoffs, ≥40% for MYC and ≥50% for BCL2 (sum < 100%), there is chance that the so-called DE comes from non-overlapping regions of the tumor (not single-cell DE). Moreover, tumor expression scores based on morphology assessment often vary among pathologists. In this study, we performed fluorescent multiplex IHC (fmIHC) using a MultiOmyxTM immunofluorescence platform (NeoGenomics Laboratories, Inc.) to determine single-cell MYC and BCL2 co-expression for prognostic analysis in DLBCL. Patients and Methods: Co-expression of MYC, BCL2, and 11 immunophenotypic markers by fmIHC was digitally quantified for 547 patients with DLBCL. Percentage of cells with MYC/BCL2 double or single expression was calculated for various cell types. Prognostic analysis was performed for de novo DLBCL cases, including 108 patients treated CHOP chemotherapy, and 356 patients treated with rituximab (R)-CHOP. Results: First, although MYC and BCL2 were expressed in multiple cell types, the vast majority (>90%) of MYC+BCL2+ DE cells were lymphoma cells. Frequencies of MYC and BCL2 co-expression were consistent with random distribution of independent MYC/BCL2 expression in positive cases. The activated B-cell-like subtype compared with the germinal center B-cell-like subtype had a significantly higher mean percentage of MYC+BCL2+ DE, but not single BCL2 or MYC expression, in PAX5+ cells. Second, in both the R-CHOP and CHOP cohorts, high percentages of DE in PAX5+ cells and in all DAPI+ cells were associated with significantly poorer survival of DLBCL patients. Very high percentage of single BCL2 expression in PAX5+ cells was also associated with poorer survival, however, only in a small number of patients in both cohorts. Only in the R-CHOP cohort, high single MYC expression in PAX5+ cells (in a significant proportion of the cohort) was associated with significantly poorer overall survival. In the R-CHOP cohort, eight cases had MYC/BCL2 double hit by FISH, including 3 cases with MYC/IGH fusions (DE) and 5 cases with MYC/non-IGH rearrangements (only one DE). The exclusion of these double hit cases did not impact the significance of DE and single MYC+ expression. Summary: True MYC+BCL2+ double protein expression at the single cell level is predominantly in lymphoma cells, and associated with significantly poor survival of DLBCL patients. Our results suggest high cutoffs for common MYC/BCL2 IHC to identify true double expressor lymphoma. Citation Format: Zijun Yidan Xu-Monette, Yong Li, Qingyan Au, Harry Nunns, Wenyu Shi, Alexandar Tzankov, Carlo Visco, Govind Bhagat, Eric Hsi, Xiaoxian X. Zhao, Karen Dybkaer, April Chiu, Wayne Tam, Youli Zu, Fredrick B. Hagemeister, Andrew Song, Michele Pellegrino, Heounjeong Go, Maurilio Ponzoni, Andrés Ferreri, Michael B. Møller, Benjamin Parsons, J. Han van Krieken, Miguel A. Piris, Jane N. Winter, Mingzhi Zhang, Bing Xu, Ken H. Young. Single-cell double expression of MYC and BCL2 proteins by fluorescent multiplex IHC is mainly in lymphoma cells with significant prognostic impact in large B-cell lymphoma while suggesting high cutoffs for routine IHC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6418.

Introduction: In recent years, increasing evidence suggests an important role of tumor-infiltrating B lymphocytes (TIL-Bs) in cancer immunity. In various types of solid tumors, high density of TIL-Bs and presence of mature tertiary lymphoid structures consistently correlate with better prognosis and response to immune checkpoint blockade therapies. Diffuse large B-cell lymphoma (DLBCL) is a type of aggressive lymphoma arising from mature B cells. TIL-Bs have been identified in DLBCL based on single-cell gene-expression data in several previous studies, but their prognostic significance in DLBCL is unknown. In this study, we determined the clinical significance of TIL-B abundance and investigated the potential underlying mechanisms. Patients and Methods: This abstract includes a discovery study in which we distinguished TIL-Bs from malignant B-cells by immunoglobulin gene (IG) clonotype analysis and a validation study using flow cytometry. In the discovery study, a molecularly well-characterized DLBCL cohort assembled by the International DLBCL R-CHOP Consortium Program was analyzed. DNA samples from DLBCL patients were sequenced by high-throughput ultra-deep sequencing assays using multiplexed forward and reverse primers. Based on the IG heavy chain (IGH) sequencing results, malignant B-cell and TIL-B VDJ clonotypes were identified for 138 patients, and the frequencies and numbers of TIL-B clonotypes in individual sequence repertoires were calculated. In 79 DLBCL cases, a DJ-only sequence was predominant and no productive VDJ sequences met the criteria for diagnostic sequences. Total frequencies of productive VDJ sequences in the sequence repertoires were then used to access high and low TIL-B frequencies. TIL-B frequencies were also evaluated by IG light chain clonotypes for these 217 patients and additional 52 patients. The TIL-B frequencies were correlated with patient survival, gene-expression profiling (GEP) data GSE31312, and frequencies of DLBCL-infiltrating immune cells quantified by fluorescent multiplex immunohistochemistry (mIHC). To validate the prognostic significance of TIL-B frequencies, we assembled a validation cohort of 94 DLBCL patients treated at Duke University Medical Center. TIL-Bs in tissues were quantified by flow cytometry immunophenotyping. Ratios of TIL-Bs to abnormal B cells were assessed for prognostic effects. Results: First, we found thathigh frequencies of TIL-Bs and clonotypic diversity were associated with significantly favorable prognostic effects in 138 DLBCL patients with their malignant B-cell IGH-VDJ clonotypes identified (Figure A). The prognostic effects of TIL-Bs outweighed those of the germinal center B-cell-like/activated B-cell-like classification (Figure B). Second, patients with high frequencies of TIL-Bs compared with those without had significant upregulation of immune response genes by GEP (Figure C), higher CD4 T cell infiltration quantified by fluorescent mIHC (Figure D), and enrichment of EcoTyper-defined specific cell states, including significantly increased frequencies of B cell state S2 (pre-memory), CD4 T cell state S3 (naïve), follicular helper T cell state S3, and endothelial cell state S3, whereas lower frequencies of B cell state S5 (plasmablast) and follicular helper T cell state S1. Third, high TIL-B abundance in 79 ‘DJ-only’ clonotype cases and additional 52 cases assessed by light chain clonotype analysis showed similar favorable prognostic effects and associations with upregulated immune gene expression and CD4 T cell infiltration. Fourth, the identified GEP signatures, prominent with genes involved in T-dependent B cell activation and T cell activation and differentiation, showed significant prognostic effects in two GEO datasets. Fifth, TIL-B abundance showed significant prognostic effects in DLBCL cases infiltrated with T-cells in both the discovery study (Figure E) and validation study (Figure F, flow cytometry cohort). Conclusions: TIL-B frequencies and clonotypic diversity have remarkable prognostic and biological effects in DLBCL, which has significant implications for developing novel therapeutic strategies. Our study identified a new prognostic biomarker in DLBCL and an important player in anti-lymphoma immune responses, providing a new point of view to analyze the tumor biology in aggressive B-cell lymphoma and precisely identify prognostic determinants.

Background: DNA mismatch repair (MMR) deficiency (dMMR), which leads to genomic and microsatellite instability, is a biomarker that predicts response to immunotherapies and has variable prognostic effects for chemotherapies in solid tumors. The dMMR can be detected using gene sequencing or immunohistochemistry for four essential MMR proteins: MSH6, MSH2, MLH1, and PMS2. Diffuse large B-cell lymphoma (DLBCL) is the most common aggressive lymphoma. Previous studies have suggested the role of MMR gene variants in DLBCL lymphomagenesis, yet the prognostic role of dMMR in DLBCL has not been well studied. Methods: We performed targeted next-generation sequencing and immunohistochemistry for MSH6, MSH2, MLH1, and PMS2 in a large cohort of DLBCL patients treated with standard chemoimmunotherapy. Gene expression profiling (GEP) was performed using the Affymetrix GeneChip Human Genome HG-U133 Plus 2.0 microarray (data in GSE31312). Fluorescent multiplex immunohistochemistry (mIHC) was performed using MultiOmyx multiplexing immunofluorescence staining protocols and antibodies against 13 immune markers. We investigated the frequencies of dMRR and MMR protein expression and correlated the mutation status and expression levels with patient survival and DLBCL biology, including the number of mutated genes, GEP data, previously analyzed DLBCL biomarkers, and immune cell immunophenotypes. Results: MMR gene mutations and loss of MMR protein expression were infrequent in DLBCL (Figure A) and did not show a significant prognostic impact. MMR proteins were commonly expressed in DLBCL samples and the germinal centers of reactive tonsil controls (Figure B), with higher mean and median percentages of tumor cells expressing MSH6 and MLH1 proteins than in the solid tumor samples used for comparison. High expression of MMR proteins was associated with Ki-67, MYC, and p53 overexpression as determined by immunohistochemistry. GEP analysis identified significantly upregulated genes involved in the mitotic cell cycle and DNA metabolism and prominent downregulated immune gene signatures in DLBCL with high MMR protein expression. Fluorescent mIHC confirmed the associations with decreased T cell abundance in MSH6/MSH2/MLH1/PMS2 highly expressing DLBCL (Figure C) independent of p53 and MYC expression status, whereas MSH6/ MLH1 mutations were associated with increased T cell frequencies. High versus low expression of MSH6, MLH1, and PMS2 showed significant unfavorable prognostic effects in DLBCL when the optimal cutoffs were used for high expression. However, when the median percentages were used as cutoffs, the prognostic effects were not significant; nonetheless, MSH6 and PMS2 high expression showed significant adverse prognostic effects in the MYC¯ DLBCL subset. Interestingly, when we used Ki-67, MYC + or BCL2 + percentages in DLBCL assessed by immunohistochemistry to subtract MMR protein percentages and then correlated the differences to survival, we found significant favorable prognostic impact by higher percentage expression differences between MMR proteins and BCL2, or by intermediately higher differences between MMR proteins and MYC. For MSH2/MLH1/PMS2, the higher percentages compared with MYC/BCL2 showed no significant correlations with T cell infiltration levels, whereas the higher differences between MSH6 and MYC/BCL2 were associated with significantly lower T cell frequencies (Figure D). Conclusions: This study revealed that high expression of MMR proteins is a common feature of DLBCL associated with lower tumor-infiltrating T cells, MYC and p53 overexpression, and context-dependent prognostic effects. In contrast, dMMR and loss of MMR proteins are infrequent and have no significant prognostic effects in DLBCL treated with standard chemoimmunotherapy. These results may have implications for understanding DLBCL biology and the low efficacy of PD-1 blockade immunotherapy in DLBCL.

Abstract Purpose: Tumor-infiltrating B lymphocytes (TIL-B) have demonstrated prognostic and predictive significance in solid cancers. In this study, we aimed to distinguish TIL-Bs from malignant B-cells in diffuse large B-cell lymphoma (DLBCL) and determine the clinical and biological significance. Experimental Design: A total of 269 patients with de novo DLBCL from the International DLBCL R-CHOP Consortium Program were studied. Ultra-deep sequencing of the immunoglobulin genes was performed to determine B-cell clonotypes. The frequencies and numbers of TIL-B clonotypes in individual repertoires were correlated with patient survival, gene expression profiling (GEP) data, and frequencies of DLBCL-infiltrating immune cells quantified by fluorescent multiplex IHC at single-cell resolution. Results: TIL-B abundance, evaluated by frequencies of normal B-cell clonotypes in the immunoglobulin repertoires, remarkably showed positive associations with significantly better survival of patients in our sequenced cohorts. DLBCLs with high versus low TIL-B abundance displayed distinct GEP signatures, increased pre-memory B-cell state and naïve CD4 T-cell state fractions, and higher CD4+ T-cell infiltration. TIL-B frequency, as a new biomarker in DLBCL, outperformed the germinal center (GC) B-cell–like/activated B-cell–like classification and TIL-T frequency. The identified TIL-B–high GEP signature, including genes upregulated during T-dependent B-cell activation and those highly expressed in normal GC B cells and T cells, showed significant favorable prognostic effects in several external validation cohorts. Conclusions: TIL-B frequency is a significant prognostic factor in DLBCL and plays a crucial role in antitumor immune responses. This study provides novel insights into the prognostic determinants in DLBCL and TIL-B functions with important therapeutic implications.

The spleen is a commonly encountered specimen in surgical pathology. However, little is known about the incidence, morphologic pattern, and clinical features of spleens involved by amyloidosis. We retrospectively identified 69 spleen amyloid cases typed using a proteomics-based method between 2008 and 2020. The frequency of amyloid types, clinicopathologic features, and distribution of amyloid deposits were assessed. Four amyloid types were detected: immunoglobulin light chain (AL) (N=30; 43.5%); leukocyte chemotactic factor 2 amyloidosis (ALECT2) (N=30; 43.5%); amyloid A (AA) (N=8; 11.6%); and fibrinogen alpha (AFib) (N=1; 1.4%). The splenic amyloid showed 5 distinct distribution patterns: (1) diffuse pattern, exhibited by most AL cases; (2) red pulp pattern, exhibited by most ALECT2 cases; (3) multinodular pattern, seen in subsets of AA and AL-kappa cases; (4) mass-forming pattern, seen in the AFib case; and (5) vascular only, seen in a subset of AA cases. Atraumatic splenic rupture was the most common reason for splenectomy in AL cases, while most ALECT2 spleens were removed incidentally during an unrelated abdominal surgery. Splenomegaly was significantly more common in AA spleens than in AL or ALECT2 spleens and was often the reason for splenectomy in this group. In conclusion, splenic amyloid may be underrecognized as it is often an incidental finding. Although, as expected, many of the spleens were involved by AL amyloidosis, ALECT2 emerged as another common spleen amyloid type. Although the spleen amyloid types exhibited characteristic distribution patterns, proteomics-based typing is warranted as some morphologic overlap still exists. Awareness of ALECT2 as a major spleen amyloid type is important for appropriate diagnostic workup and patient management.