Abstract
Background: Tumor-specific HLA-ABC expression is required for cytotoxic T-cell elimination of cancer cells expressing tumor associated- or neo-antigens. Moreover, the success of MHC-I-targeted cancer vaccines typically hinge on the carriage of specific types (e.g. HLA-A2) in the patient as well as type-specific antigen presentation in the tumor itself. In contrast to specific type carriage, the retention of protein expression of individual HLA genes (e.g. A/B/C) is not assessed in situ in tumor cells, despite the widely recognized knowledge that cancers downregulate or eliminate antigen presentation via multifaceted mechanisms to avoid adaptive immunity. Unfortunately, the highly polymorphic nature of the genes encoding these proteins coupled with quaternary-structure changes after formalin-fixation, complicate detection by immunohistochemistry. Moreover, only a few commercially available antibodies recommended for IHC have been characterized to determine their recognition of specific HLA-ABC alleles.
Results: In this study, we determined allele and type-specificity for 15 commercially available antibodies for immunohistochemical use by staining HLA-ABC-null K562 cells transduced with 16 specific common HLA-A, B and C alleles. We identified and validated EMR8-5, EPR22172 and HC-10 as true pan-HLA-ABC antibodies and EPR1394Y, 2A11G7, EPR26121-73 as specific HLA-A, B, and C antibodies, respectively, to facilitate the investigation of HLA-ABC loss in situ. We applied this novel approach to a series of early-stage HER2-negative breast cancers (n=175) and DCIS (n=24) as a proof of utility. HLA-A, B and/or C expression on <15% of tumor cells was defined as loss, 15-75% as partial loss and >75% as fully conserved expression. Samples from hormone receptor-positive (HR+) breast cancer patients showed a higher rate of total HLA-A, B and/or C loss, any combination, than triple-negative breast cancer (TNBC) patient samples (73% vs 50% Fishers test p=0.0027, n=89 vs 77), driven by loss of HLA-A and/or B (total HLA-A loss 69% vs 30% p=0.0001 and total HLA-B loss 59% vs 34% p=0.0048). In contrast, samples from TNBC patients showed a higher rate of conserved expression for each single molecule and combined HLA-ABC (total conserved HLA-ABC 2% vs 16% in HR+HER2- vs TNBC p=0.0009). Among cases with HLA-A, B and/or C expression, most showed partial loss of at least one molecule, partial expression/loss was often observed within matching discrete spatial areas for HLA-A, B and/or C. Samples from patients with DCIS showed a pattern of HLA-A, B and/or C loss more similar to HR+HER2- invasive cancer regardless of HR status, both showing prominent AB loss and no cases with single molecule C loss. DCIS presented lower rates of HLA-C loss, any combination, than invasive carcinoma (8% vs 32% p=0.0165) regardless of subtype. Of note, the pan-HLA-ABC clones were insufficient to identify single or double molecule loss. All samples showed HLA-ABC expression on surrounding non-tumor tissue; however, we observed lack of staining on tumor and non-tumor cells in 5 samples with HLA-C specific EPR26121-73 antibody that were excluded from the analysis.
Discussion: We demonstrate the novel finding that up to 70% of early HR+ breast cancers lack any detectable HLA-A expression, compared to 30% of early triple-negative breast cancers. HLA-A and B loss are early events present in DCIS while HLA-C loss was more common in invasive carcinoma. Moreover, HLA-C expression was the most conserved across breast cancer subtypes. We also report an unprecedented level of individual gene product expression loss across breast cancers, which would not be detectable using poorly validated or pan-MHC-I antibodies in situ. These findings have clear implications on the success of checkpoint inhibitors and vaccine strategies in this setting, and hint at ‘para-genomic’ mechanisms of MHC-I gene-specific silencing in cancer-specific contexts, enabling the discovery of novel tumor evolution principles.
Results: In this study, we determined allele and type-specificity for 15 commercially available antibodies for immunohistochemical use by staining HLA-ABC-null K562 cells transduced with 16 specific common HLA-A, B and C alleles. We identified and validated EMR8-5, EPR22172 and HC-10 as true pan-HLA-ABC antibodies and EPR1394Y, 2A11G7, EPR26121-73 as specific HLA-A, B, and C antibodies, respectively, to facilitate the investigation of HLA-ABC loss in situ. We applied this novel approach to a series of early-stage HER2-negative breast cancers (n=175) and DCIS (n=24) as a proof of utility. HLA-A, B and/or C expression on <15% of tumor cells was defined as loss, 15-75% as partial loss and >75% as fully conserved expression. Samples from hormone receptor-positive (HR+) breast cancer patients showed a higher rate of total HLA-A, B and/or C loss, any combination, than triple-negative breast cancer (TNBC) patient samples (73% vs 50% Fishers test p=0.0027, n=89 vs 77), driven by loss of HLA-A and/or B (total HLA-A loss 69% vs 30% p=0.0001 and total HLA-B loss 59% vs 34% p=0.0048). In contrast, samples from TNBC patients showed a higher rate of conserved expression for each single molecule and combined HLA-ABC (total conserved HLA-ABC 2% vs 16% in HR+HER2- vs TNBC p=0.0009). Among cases with HLA-A, B and/or C expression, most showed partial loss of at least one molecule, partial expression/loss was often observed within matching discrete spatial areas for HLA-A, B and/or C. Samples from patients with DCIS showed a pattern of HLA-A, B and/or C loss more similar to HR+HER2- invasive cancer regardless of HR status, both showing prominent AB loss and no cases with single molecule C loss. DCIS presented lower rates of HLA-C loss, any combination, than invasive carcinoma (8% vs 32% p=0.0165) regardless of subtype. Of note, the pan-HLA-ABC clones were insufficient to identify single or double molecule loss. All samples showed HLA-ABC expression on surrounding non-tumor tissue; however, we observed lack of staining on tumor and non-tumor cells in 5 samples with HLA-C specific EPR26121-73 antibody that were excluded from the analysis.
Discussion: We demonstrate the novel finding that up to 70% of early HR+ breast cancers lack any detectable HLA-A expression, compared to 30% of early triple-negative breast cancers. HLA-A and B loss are early events present in DCIS while HLA-C loss was more common in invasive carcinoma. Moreover, HLA-C expression was the most conserved across breast cancer subtypes. We also report an unprecedented level of individual gene product expression loss across breast cancers, which would not be detectable using poorly validated or pan-MHC-I antibodies in situ. These findings have clear implications on the success of checkpoint inhibitors and vaccine strategies in this setting, and hint at ‘para-genomic’ mechanisms of MHC-I gene-specific silencing in cancer-specific contexts, enabling the discovery of novel tumor evolution principles.
Background: Tumor-specific HLA-ABC expression is required for cytotoxic T-cell elimination of cancer cells expressing tumor associated- or neo-antigens. Moreover, the success of MHC-I-targeted cancer vaccines typically hinge on the carriage of specific types (e.g. HLA-A2) in the patient as well as type-specific antigen presentation in the tumor itself. In contrast to specific type carriage, the retention of protein expression of individual HLA genes (e.g. A/B/C) is not assessed in situ in tumor cells, despite the widely recognized knowledge that cancers downregulate or eliminate antigen presentation via multifaceted mechanisms to avoid adaptive immunity. Unfortunately, the highly polymorphic nature of the genes encoding these proteins coupled with quaternary-structure changes after formalin-fixation, complicate detection by immunohistochemistry. Moreover, only a few commercially available antibodies recommended for IHC have been characterized to determine their recognition of specific HLA-ABC alleles.
Results: In this study, we determined allele and type-specificity for 15 commercially available antibodies for immunohistochemical use by staining HLA-ABC-null K562 cells transduced with 16 specific common HLA-A, B and C alleles. We identified and validated EMR8-5, EPR22172 and HC-10 as true pan-HLA-ABC antibodies and EPR1394Y, 2A11G7, EPR26121-73 as specific HLA-A, B, and C antibodies, respectively, to facilitate the investigation of HLA-ABC loss in situ. We applied this novel approach to a series of early-stage HER2-negative breast cancers (n=175) and DCIS (n=24) as a proof of utility. HLA-A, B and/or C expression on <15% of tumor cells was defined as loss, 15-75% as partial loss and >75% as fully conserved expression. Samples from hormone receptor-positive (HR+) breast cancer patients showed a higher rate of total HLA-A, B and/or C loss, any combination, than triple-negative breast cancer (TNBC) patient samples (73% vs 50% Fishers test p=0.0027, n=89 vs 77), driven by loss of HLA-A and/or B (total HLA-A loss 69% vs 30% p=0.0001 and total HLA-B loss 59% vs 34% p=0.0048). In contrast, samples from TNBC patients showed a higher rate of conserved expression for each single molecule and combined HLA-ABC (total conserved HLA-ABC 2% vs 16% in HR+HER2- vs TNBC p=0.0009). Among cases with HLA-A, B and/or C expression, most showed partial loss of at least one molecule, partial expression/loss was often observed within matching discrete spatial areas for HLA-A, B and/or C. Samples from patients with DCIS showed a pattern of HLA-A, B and/or C loss more similar to HR+HER2- invasive cancer regardless of HR status, both showing prominent AB loss and no cases with single molecule C loss. DCIS presented lower rates of HLA-C loss, any combination, than invasive carcinoma (8% vs 32% p=0.0165) regardless of subtype. Of note, the pan-HLA-ABC clones were insufficient to identify single or double molecule loss. All samples showed HLA-ABC expression on surrounding non-tumor tissue; however, we observed lack of staining on tumor and non-tumor cells in 5 samples with HLA-C specific EPR26121-73 antibody that were excluded from the analysis.
Discussion: We demonstrate the novel finding that up to 70% of early HR+ breast cancers lack any detectable HLA-A expression, compared to 30% of early triple-negative breast cancers. HLA-A and B loss are early events present in DCIS while HLA-C loss was more common in invasive carcinoma. Moreover, HLA-C expression was the most conserved across breast cancer subtypes. We also report an unprecedented level of individual gene product expression loss across breast cancers, which would not be detectable using poorly validated or pan-MHC-I antibodies in situ. These findings have clear implications on the success of checkpoint inhibitors and vaccine strategies in this setting, and hint at ‘para-genomic’ mechanisms of MHC-I gene-specific silencing in cancer-specific contexts, enabling the discovery of novel tumor evolution principles.
Results: In this study, we determined allele and type-specificity for 15 commercially available antibodies for immunohistochemical use by staining HLA-ABC-null K562 cells transduced with 16 specific common HLA-A, B and C alleles. We identified and validated EMR8-5, EPR22172 and HC-10 as true pan-HLA-ABC antibodies and EPR1394Y, 2A11G7, EPR26121-73 as specific HLA-A, B, and C antibodies, respectively, to facilitate the investigation of HLA-ABC loss in situ. We applied this novel approach to a series of early-stage HER2-negative breast cancers (n=175) and DCIS (n=24) as a proof of utility. HLA-A, B and/or C expression on <15% of tumor cells was defined as loss, 15-75% as partial loss and >75% as fully conserved expression. Samples from hormone receptor-positive (HR+) breast cancer patients showed a higher rate of total HLA-A, B and/or C loss, any combination, than triple-negative breast cancer (TNBC) patient samples (73% vs 50% Fishers test p=0.0027, n=89 vs 77), driven by loss of HLA-A and/or B (total HLA-A loss 69% vs 30% p=0.0001 and total HLA-B loss 59% vs 34% p=0.0048). In contrast, samples from TNBC patients showed a higher rate of conserved expression for each single molecule and combined HLA-ABC (total conserved HLA-ABC 2% vs 16% in HR+HER2- vs TNBC p=0.0009). Among cases with HLA-A, B and/or C expression, most showed partial loss of at least one molecule, partial expression/loss was often observed within matching discrete spatial areas for HLA-A, B and/or C. Samples from patients with DCIS showed a pattern of HLA-A, B and/or C loss more similar to HR+HER2- invasive cancer regardless of HR status, both showing prominent AB loss and no cases with single molecule C loss. DCIS presented lower rates of HLA-C loss, any combination, than invasive carcinoma (8% vs 32% p=0.0165) regardless of subtype. Of note, the pan-HLA-ABC clones were insufficient to identify single or double molecule loss. All samples showed HLA-ABC expression on surrounding non-tumor tissue; however, we observed lack of staining on tumor and non-tumor cells in 5 samples with HLA-C specific EPR26121-73 antibody that were excluded from the analysis.
Discussion: We demonstrate the novel finding that up to 70% of early HR+ breast cancers lack any detectable HLA-A expression, compared to 30% of early triple-negative breast cancers. HLA-A and B loss are early events present in DCIS while HLA-C loss was more common in invasive carcinoma. Moreover, HLA-C expression was the most conserved across breast cancer subtypes. We also report an unprecedented level of individual gene product expression loss across breast cancers, which would not be detectable using poorly validated or pan-MHC-I antibodies in situ. These findings have clear implications on the success of checkpoint inhibitors and vaccine strategies in this setting, and hint at ‘para-genomic’ mechanisms of MHC-I gene-specific silencing in cancer-specific contexts, enabling the discovery of novel tumor evolution principles.
In situ detection of individual classical MHC-I gene products in breast cancer identifies gene- and subtype-specific biased antigen presentation loss
Paula Gonzalez-Ericsson
Abstract
Background: Tumor-specific HLA-ABC expression is required for cytotoxic T-cell elimination of cancer cells expressing tumor associated- or neo-antigens. Moreover, the success of MHC-I-targeted cancer vaccines typically hinge on the carriage of specific types (e.g. HLA-A2) in the patient as well as type-specific antigen presentation in the tumor itself. In contrast to specific type carriage, the retention of protein expression of individual HLA genes (e.g. A/B/C) is not assessed in situ in tumor cells, despite the widely recognized knowledge that cancers downregulate or eliminate antigen presentation via multifaceted mechanisms to avoid adaptive immunity. Unfortunately, the highly polymorphic nature of the genes encoding these proteins coupled with quaternary-structure changes after formalin-fixation, complicate detection by immunohistochemistry. Moreover, only a few commercially available antibodies recommended for IHC have been characterized to determine their recognition of specific HLA-ABC alleles.
Results: In this study, we determined allele and type-specificity for 15 commercially available antibodies for immunohistochemical use by staining HLA-ABC-null K562 cells transduced with 16 specific common HLA-A, B and C alleles. We identified and validated EMR8-5, EPR22172 and HC-10 as true pan-HLA-ABC antibodies and EPR1394Y, 2A11G7, EPR26121-73 as specific HLA-A, B, and C antibodies, respectively, to facilitate the investigation of HLA-ABC loss in situ. We applied this novel approach to a series of early-stage HER2-negative breast cancers (n=175) and DCIS (n=24) as a proof of utility. HLA-A, B and/or C expression on <15% of tumor cells was defined as loss, 15-75% as partial loss and >75% as fully conserved expression. Samples from hormone receptor-positive (HR+) breast cancer patients showed a higher rate of total HLA-A, B and/or C loss, any combination, than triple-negative breast cancer (TNBC) patient samples (73% vs 50% Fishers test p=0.0027, n=89 vs 77), driven by loss of HLA-A and/or B (total HLA-A loss 69% vs 30% p=0.0001 and total HLA-B loss 59% vs 34% p=0.0048). In contrast, samples from TNBC patients showed a higher rate of conserved expression for each single molecule and combined HLA-ABC (total conserved HLA-ABC 2% vs 16% in HR+HER2- vs TNBC p=0.0009). Among cases with HLA-A, B and/or C expression, most showed partial loss of at least one molecule, partial expression/loss was often observed within matching discrete spatial areas for HLA-A, B and/or C. Samples from patients with DCIS showed a pattern of HLA-A, B and/or C loss more similar to HR+HER2- invasive cancer regardless of HR status, both showing prominent AB loss and no cases with single molecule C loss. DCIS presented lower rates of HLA-C loss, any combination, than invasive carcinoma (8% vs 32% p=0.0165) regardless of subtype. Of note, the pan-HLA-ABC clones were insufficient to identify single or double molecule loss. All samples showed HLA-ABC expression on surrounding non-tumor tissue; however, we observed lack of staining on tumor and non-tumor cells in 5 samples with HLA-C specific EPR26121-73 antibody that were excluded from the analysis.
Discussion: We demonstrate the novel finding that up to 70% of early HR+ breast cancers lack any detectable HLA-A expression, compared to 30% of early triple-negative breast cancers. HLA-A and B loss are early events present in DCIS while HLA-C loss was more common in invasive carcinoma. Moreover, HLA-C expression was the most conserved across breast cancer subtypes. We also report an unprecedented level of individual gene product expression loss across breast cancers, which would not be detectable using poorly validated or pan-MHC-I antibodies in situ. These findings have clear implications on the success of checkpoint inhibitors and vaccine strategies in this setting, and hint at ‘para-genomic’ mechanisms of MHC-I gene-specific silencing in cancer-specific contexts, enabling the discovery of novel tumor evolution principles.
Results: In this study, we determined allele and type-specificity for 15 commercially available antibodies for immunohistochemical use by staining HLA-ABC-null K562 cells transduced with 16 specific common HLA-A, B and C alleles. We identified and validated EMR8-5, EPR22172 and HC-10 as true pan-HLA-ABC antibodies and EPR1394Y, 2A11G7, EPR26121-73 as specific HLA-A, B, and C antibodies, respectively, to facilitate the investigation of HLA-ABC loss in situ. We applied this novel approach to a series of early-stage HER2-negative breast cancers (n=175) and DCIS (n=24) as a proof of utility. HLA-A, B and/or C expression on <15% of tumor cells was defined as loss, 15-75% as partial loss and >75% as fully conserved expression. Samples from hormone receptor-positive (HR+) breast cancer patients showed a higher rate of total HLA-A, B and/or C loss, any combination, than triple-negative breast cancer (TNBC) patient samples (73% vs 50% Fishers test p=0.0027, n=89 vs 77), driven by loss of HLA-A and/or B (total HLA-A loss 69% vs 30% p=0.0001 and total HLA-B loss 59% vs 34% p=0.0048). In contrast, samples from TNBC patients showed a higher rate of conserved expression for each single molecule and combined HLA-ABC (total conserved HLA-ABC 2% vs 16% in HR+HER2- vs TNBC p=0.0009). Among cases with HLA-A, B and/or C expression, most showed partial loss of at least one molecule, partial expression/loss was often observed within matching discrete spatial areas for HLA-A, B and/or C. Samples from patients with DCIS showed a pattern of HLA-A, B and/or C loss more similar to HR+HER2- invasive cancer regardless of HR status, both showing prominent AB loss and no cases with single molecule C loss. DCIS presented lower rates of HLA-C loss, any combination, than invasive carcinoma (8% vs 32% p=0.0165) regardless of subtype. Of note, the pan-HLA-ABC clones were insufficient to identify single or double molecule loss. All samples showed HLA-ABC expression on surrounding non-tumor tissue; however, we observed lack of staining on tumor and non-tumor cells in 5 samples with HLA-C specific EPR26121-73 antibody that were excluded from the analysis.
Discussion: We demonstrate the novel finding that up to 70% of early HR+ breast cancers lack any detectable HLA-A expression, compared to 30% of early triple-negative breast cancers. HLA-A and B loss are early events present in DCIS while HLA-C loss was more common in invasive carcinoma. Moreover, HLA-C expression was the most conserved across breast cancer subtypes. We also report an unprecedented level of individual gene product expression loss across breast cancers, which would not be detectable using poorly validated or pan-MHC-I antibodies in situ. These findings have clear implications on the success of checkpoint inhibitors and vaccine strategies in this setting, and hint at ‘para-genomic’ mechanisms of MHC-I gene-specific silencing in cancer-specific contexts, enabling the discovery of novel tumor evolution principles.
Background: Tumor-specific HLA-ABC expression is required for cytotoxic T-cell elimination of cancer cells expressing tumor associated- or neo-antigens. Moreover, the success of MHC-I-targeted cancer vaccines typically hinge on the carriage of specific types (e.g. HLA-A2) in the patient as well as type-specific antigen presentation in the tumor itself. In contrast to specific type carriage, the retention of protein expression of individual HLA genes (e.g. A/B/C) is not assessed in situ in tumor cells, despite the widely recognized knowledge that cancers downregulate or eliminate antigen presentation via multifaceted mechanisms to avoid adaptive immunity. Unfortunately, the highly polymorphic nature of the genes encoding these proteins coupled with quaternary-structure changes after formalin-fixation, complicate detection by immunohistochemistry. Moreover, only a few commercially available antibodies recommended for IHC have been characterized to determine their recognition of specific HLA-ABC alleles.
Results: In this study, we determined allele and type-specificity for 15 commercially available antibodies for immunohistochemical use by staining HLA-ABC-null K562 cells transduced with 16 specific common HLA-A, B and C alleles. We identified and validated EMR8-5, EPR22172 and HC-10 as true pan-HLA-ABC antibodies and EPR1394Y, 2A11G7, EPR26121-73 as specific HLA-A, B, and C antibodies, respectively, to facilitate the investigation of HLA-ABC loss in situ. We applied this novel approach to a series of early-stage HER2-negative breast cancers (n=175) and DCIS (n=24) as a proof of utility. HLA-A, B and/or C expression on <15% of tumor cells was defined as loss, 15-75% as partial loss and >75% as fully conserved expression. Samples from hormone receptor-positive (HR+) breast cancer patients showed a higher rate of total HLA-A, B and/or C loss, any combination, than triple-negative breast cancer (TNBC) patient samples (73% vs 50% Fishers test p=0.0027, n=89 vs 77), driven by loss of HLA-A and/or B (total HLA-A loss 69% vs 30% p=0.0001 and total HLA-B loss 59% vs 34% p=0.0048). In contrast, samples from TNBC patients showed a higher rate of conserved expression for each single molecule and combined HLA-ABC (total conserved HLA-ABC 2% vs 16% in HR+HER2- vs TNBC p=0.0009). Among cases with HLA-A, B and/or C expression, most showed partial loss of at least one molecule, partial expression/loss was often observed within matching discrete spatial areas for HLA-A, B and/or C. Samples from patients with DCIS showed a pattern of HLA-A, B and/or C loss more similar to HR+HER2- invasive cancer regardless of HR status, both showing prominent AB loss and no cases with single molecule C loss. DCIS presented lower rates of HLA-C loss, any combination, than invasive carcinoma (8% vs 32% p=0.0165) regardless of subtype. Of note, the pan-HLA-ABC clones were insufficient to identify single or double molecule loss. All samples showed HLA-ABC expression on surrounding non-tumor tissue; however, we observed lack of staining on tumor and non-tumor cells in 5 samples with HLA-C specific EPR26121-73 antibody that were excluded from the analysis.
Discussion: We demonstrate the novel finding that up to 70% of early HR+ breast cancers lack any detectable HLA-A expression, compared to 30% of early triple-negative breast cancers. HLA-A and B loss are early events present in DCIS while HLA-C loss was more common in invasive carcinoma. Moreover, HLA-C expression was the most conserved across breast cancer subtypes. We also report an unprecedented level of individual gene product expression loss across breast cancers, which would not be detectable using poorly validated or pan-MHC-I antibodies in situ. These findings have clear implications on the success of checkpoint inhibitors and vaccine strategies in this setting, and hint at ‘para-genomic’ mechanisms of MHC-I gene-specific silencing in cancer-specific contexts, enabling the discovery of novel tumor evolution principles.
Results: In this study, we determined allele and type-specificity for 15 commercially available antibodies for immunohistochemical use by staining HLA-ABC-null K562 cells transduced with 16 specific common HLA-A, B and C alleles. We identified and validated EMR8-5, EPR22172 and HC-10 as true pan-HLA-ABC antibodies and EPR1394Y, 2A11G7, EPR26121-73 as specific HLA-A, B, and C antibodies, respectively, to facilitate the investigation of HLA-ABC loss in situ. We applied this novel approach to a series of early-stage HER2-negative breast cancers (n=175) and DCIS (n=24) as a proof of utility. HLA-A, B and/or C expression on <15% of tumor cells was defined as loss, 15-75% as partial loss and >75% as fully conserved expression. Samples from hormone receptor-positive (HR+) breast cancer patients showed a higher rate of total HLA-A, B and/or C loss, any combination, than triple-negative breast cancer (TNBC) patient samples (73% vs 50% Fishers test p=0.0027, n=89 vs 77), driven by loss of HLA-A and/or B (total HLA-A loss 69% vs 30% p=0.0001 and total HLA-B loss 59% vs 34% p=0.0048). In contrast, samples from TNBC patients showed a higher rate of conserved expression for each single molecule and combined HLA-ABC (total conserved HLA-ABC 2% vs 16% in HR+HER2- vs TNBC p=0.0009). Among cases with HLA-A, B and/or C expression, most showed partial loss of at least one molecule, partial expression/loss was often observed within matching discrete spatial areas for HLA-A, B and/or C. Samples from patients with DCIS showed a pattern of HLA-A, B and/or C loss more similar to HR+HER2- invasive cancer regardless of HR status, both showing prominent AB loss and no cases with single molecule C loss. DCIS presented lower rates of HLA-C loss, any combination, than invasive carcinoma (8% vs 32% p=0.0165) regardless of subtype. Of note, the pan-HLA-ABC clones were insufficient to identify single or double molecule loss. All samples showed HLA-ABC expression on surrounding non-tumor tissue; however, we observed lack of staining on tumor and non-tumor cells in 5 samples with HLA-C specific EPR26121-73 antibody that were excluded from the analysis.
Discussion: We demonstrate the novel finding that up to 70% of early HR+ breast cancers lack any detectable HLA-A expression, compared to 30% of early triple-negative breast cancers. HLA-A and B loss are early events present in DCIS while HLA-C loss was more common in invasive carcinoma. Moreover, HLA-C expression was the most conserved across breast cancer subtypes. We also report an unprecedented level of individual gene product expression loss across breast cancers, which would not be detectable using poorly validated or pan-MHC-I antibodies in situ. These findings have clear implications on the success of checkpoint inhibitors and vaccine strategies in this setting, and hint at ‘para-genomic’ mechanisms of MHC-I gene-specific silencing in cancer-specific contexts, enabling the discovery of novel tumor evolution principles.
{{ help_message }}
{{filter}}