Search

To determine the prevalence of red blood cell (RBC) alloimmunisation and alloantibody specificity in sickle cell disease (SCD) patients in Kisangani, Democratic Republic of Congo (DRC) in comparison with those followed at the Centre Hospitalier Régional (CHR) de la Citadelle of Liège (Belgium).Data regarding RBC alloimmunisation (immune response of the organism to foreign erythrocyte antigens, antigens that lack on its own RBC) in SCD patients are scarce in sub-Saharan Africa.We conducted a multi-site-based cross-sectional study among 125 SCD patients at Kisangani and 136 at the CHR de la Citadelle of Liège. The diagnosis of SCD was confirmed by high-performance liquid chromatography. Alloantibodies were screened using the agglutination technique on gel cards and their specificity determined using 11 and/or 16 cell panels. Statistical analyses were carried out using SPSS.The prevalence of RBC alloimmunisation was 9.6% among SCD patients in Kisangani versus 22.8% in those of Liège. At Kisangani as well as at Liège, the median age of alloimmunised patients was higher than that of non-alloimmunised patients, 15.5 years (IQR:4.8-19.8) and 24 years (IQR:14-31) versus 10 years (IQR: 6.5-17) and 17 years (IQR:12-24), respectively. The median number of blood units was higher in both Kisangani and Liège immunised patients compared to non-immunised patients, 8 (IQR:5-11) versus 5 (IQR:3-13) and 41(IQR:6-93) versus 6.5(3-37) respectively. At Kisangani (N = 14), the most frequent antibodies were anti-D (28.6%) and anti-C versus anti-E (13.6%), anti-S (13.6%) and anti-Lea (11.4%) at Liège (N = 44).These findings stated that alloimmunisation is a common complication in SCD patients in the DRC. In the resource-limited setting of this country, blood transfusion with minimal ABO, D, C and E antigen matching in addition to the use of compatibility test could significantly reduce the incidence of this complication.

Recent advances in high-throughput technologies enable a more comprehensive interpretation of the tumour evolution through the study of the intra-tumour heterogeneity. Several algorithms, however, often relies on the use of models that described the top of the iceberg regarding the stromal contamination of the samples, making diagnosis difficult to assess. Indeed, such as radio wave receivers, tools to analyse high-throughput technologies data, are used to enable the discrimination between multiple signals differing in frequencies. However, such tools often look at the average frequency more than distinct signals, leading to analyse a confused signal. This confusion could dramatically lead to a mis–interpretation of the real data, especially during the diagnosis as it relies on the choice of a unique scenario among many others. Here, we describe how this signal confusion occurs in the most classical DNA microarray analysis of tumours and we provide statistics to determine how many other possible scenario can lead the same signals, in order to improve the robustness of pigeon hole logic based analysis. Based on simulations, where a unique tumour population was diluted by an increasing gradient of normal cells, we underline the causes and consequences of such signal confusion for up to five allelic copies. Despite the removal of all technical biaises and background noise, we show how the signal confusion remains systematically present in the commonly used DNA microarray analysis, especially for the genotypes AAAAB, AAAB and AAB for copy numbers 5, 4 and 3 respectively, as well as their symmetric combinations for the B allele.

Biomarker signature discovery remains the main path to develop clinical diagnostic tools when the biological knowledge on a pathology is weak. Shortest signatures are often preferred to reduce the cost of the diagnostic. The ability to find the best and shortest signature relies on the robustness of the models that can be built on such set of molecules. The classification algorithm that will be used is selected based on the average performance of its models, often expressed via the average AUC. However, it is not garanteed that an algorithm with a large AUC distribution will keep a stable performance when facing data. Here, we propose two AUC-derived hyper-stability scores, the HRS and the HSS, as complementary metrics to the average AUC, that should bring confidence in the choice for the best classification algorithm. To emphasize the importance of these scores, we compared 15 different Random Forests implementation. Additionally, the modelization time of each implementation was computed to further help deciding the best strategy. Our findings show that the Random Forest implementation should be chosen according to the data at hand and the classification question being evaluated. No Random Forest implementation can be used universally for any classification and on any dataset. Each of them should be tested for both their average AUC performance and AUC-derived stability, prior to analysis.Author summaryTo better measure the performance of a Machine Learning (ML) implementation, we introduce a new metric, the AUC hyper-stability, to be used in parallel with the average AUC. This AUC hyper-stability is able to discriminate ML implementations that show the same AUC performance. This metric can therefore help researchers in choosing the best ML method to get stable short predictive biomarker signatures. More specifically, we advocate a tradeoff between the average AUC performance, the hyper-stability scores, and the modeling time.

Rapid Whole Genome Sequencing (rWGS) represents a valuable exploration in critically ill pediatric patients. Early diagnosis allows care to be adjusted. We evaluated the feasibility, turnaround time (TAT), yield, and utility of rWGS in Belgium. Twenty-one unrelated critically ill patients were recruited from the neonatal intensive care units, the pediatric intensive care unit, and the neuropediatric unit, and offered rWGS as a first tier test. Libraries were prepared in the laboratory of human genetics of the University of Liège using Illumina DNA PCR-free protocol. Sequencing was performed on a NovaSeq 6000 in trio for 19 and in duo for two probands. The TAT was calculated from the sample reception to the validation of results. Clinical utility data were provided by treating physicians. A definite diagnosis was reached in twelve (57.5%) patients in 39.80 h on average (range: 37.05–43.7). An unsuspected diagnosis was identified in seven patients. rWGS guided care adjustments in diagnosed patients, including a gene therapy, an off-label drug trial and two condition-specific treatments. We successfully implemented the fastest rWGS platform in Europe and obtained one of the highest rWGS yields. This study establishes the path for a nationwide semi-centered rWGS network in Belgium.

Infertility in couples is a common problem, with both female and male factors contributing to similar extents. Severe, congenital disorders affecting fertility are, however, rare. While folliculogenesis and spermatogenesis are generally orchestrated via different mechanisms, some genetic anomalies can impair both female and male gametogenesis. Minichromosome maintenance complex component 9 (MCM9) is involved in DNA repair and mutations of the MCM9 gene have been previously reported in females with premature ovarian insufficiency (POI). MCM9 is also an emerging cancer risk gene. We performed next-generation and Sanger sequencing of fertility and related genes and hormonal and imaging studies in a kindred whose members had POI and disordered spermatogenesis. We identified a homozygous pathogenic MCM9 variant, c.394C>T (p.Arg132*) in three sisters affected by POI due to ovarian dysgenesis and their brother who had normal pubertal development but suffered from non-obstructive azoospermia. Testicular biopsy revealed Sertoli cell-only testicular histopathology. No evidence of early onset cancer was found in the homozygotic family members, but they were all young (

Genotype; Cancer phenotypes; Genetic tumour Genotip; Fenotips del càncer; Tumor genètic Genotipo; Fenotipos del cáncer; Tumor genético Background Truncating pathogenic or likely pathogenic variants of CDH1 cause hereditary diffuse gastric cancer (HDGC), a tumour risk syndrome that predisposes carrier individuals to diffuse gastric and lobular breast cancer. Rare CDH1 missense variants are often classified as variants of unknown significance. We conducted a genotype–phenotype analysis in families carrying rare CDH1 variants, comparing cancer spectrum in carriers of pathogenic or likely pathogenic variants (PV/LPV; analysed jointly) or missense variants of unknown significance, assessing the frequency of families with lobular breast cancer among PV/LPV carrier families, and testing the performance of lobular breast cancer-expanded criteria for CDH1 testing. Methods This genotype-first study used retrospective diagnostic and clinical data from 854 carriers of 398 rare CDH1 variants and 1021 relatives, irrespective of HDGC clinical criteria, from 29 institutions in ten member-countries of the European Reference Network on Tumour Risk Syndromes (ERN GENTURIS). Data were collected from Oct 1, 2018, to Sept 20, 2022. Variants were classified by molecular type and clinical actionability with the American College of Medical Genetics and Association for Molecular Pathology CDH1 guidelines (version 2). Families were categorised by whether they fulfilled the 2015 and 2020 HDGC clinical criteria. Genotype–phenotype associations were analysed by Student's t test, Kruskal-Wallis, χ2, and multivariable logistic regression models. Performance of HDGC clinical criteria sets were assessed with an equivalence test and Youden index, and the areas under the receiver operating characteristic curves were compared by Z test. Findings From 1971 phenotypes (contributed by 854 probands and 1021 relatives aged 1–93 years), 460 had gastric and breast cancer histology available. CDH1 truncating PV/LPVs occurred in 176 (21%) of 854 families and missense variants of unknown significance in 169 (20%) families. Multivariable logistic regression comparing phenotypes occurring in families carrying PV/LPVs or missense variants of unknown significance showed that lobular breast cancer had the greatest positive association with the presence of PV/LPVs (odds ratio 12·39 [95% CI 2·66–57·74], p=0·0014), followed by diffuse gastric cancer (8·00 [2·18–29·39], p=0·0017) and gastric cancer (7·81 [2·03–29·96], p=0·0027). 136 (77%) of 176 families carrying PV/LPVs fulfilled the 2015 HDGC criteria. Of the remaining 40 (23%) families, who did not fulfil the 2015 criteria, 11 fulfilled the 2020 HDGC criteria, and 18 had lobular breast cancer only or lobular breast cancer and gastric cancer, but did not meet the 2020 criteria. No specific CDH1 variant was found to predispose individuals specifically to lobular breast cancer, although 12 (7%) of 176 PV/LPV carrier families had lobular breast cancer only. Addition of three new lobular breast cancer-centred criteria improved testing sensitivity while retaining high specificity. The probability of finding CDH1 PV/LPVs in patients fulfilling the lobular breast cancer-expanded criteria, compared with the 2020 criteria, increased significantly (AUC 0·92 vs 0·88; Z score 3·54; p=0·0004). Interpretation CDH1 PV/LPVs were positively associated with HDGC-related phenotypes (lobular breast cancer, diffuse gastric cancer, and gastric cancer), and no evidence for a positive association with these phenotypes was found for CDH1 missense variants of unknown significance. CDH1 PV/LPVs occurred often in families with lobular breast cancer who did not fulfil the 2020 HDGC criteria, supporting the expansion of lobular breast cancer-centred criteria. Funding European Reference Network on Genetic Tumour Risk Syndromes, European Regional Development Fund, Fundação para a Ciência e a Tecnologia (Portugal), Cancer Research UK, and European Union's Horizon 2020 research and innovation programme. European Reference Network on Genetic Tumour Risk Syndromes, European Regional Development Fund, Fundação para a Ciência e a Tecnologia (Portugal), Cancer Research UK, and European Union's Horizon 2020 research and innovation programme.

Mathematical modelling studies have shown that repetitive screening can be used to mitigate SARS-CoV-2 transmission in primary schools while keeping schools open. However, not much is known about how transmission progresses within schools and whether there is a risk of importation to households. During the academic year 2020-2021, a prospective surveillance study using repetitive screening was conducted in a primary school and associated households in Liège (Belgium). SARS-CoV-2 screening was performed via throat washing either once or twice a week. We used genomic and epidemiological data to reconstruct the observed school outbreaks using two different models. The outbreaker2 model combines information on the generation time and contact patterns with a model of sequence evolution. For comparison we also used SCOTTI, a phylogenetic model based on the structured coalescent. In addition, we performed a simulation study to investigate how the accuracy of estimated positivity rates in a school depends on the proportion of a school that is sampled in a repetitive screening strategy. We found no difference in SARS-CoV-2 positivity between children and adults and children were not more often asymptomatic compared to adults. Both models for outbreak reconstruction revealed that transmission occurred mainly within the school environment. Uncertainty in outbreak reconstruction was lowest when including genomic as well as epidemiological data. We found that observed weekly positivity rates are a good approximation to the true weekly positivity rate, especially in children, even when only 25% of the school population is sampled. These results indicate that, in addition to reducing infections as shown in modelling studies, repetitive screening in school settings can lead to a better understanding of the extent of transmission in schools during a pandemic and importation risk at the community level. ispartof: Epidemics vol:44 pages:100701- ispartof: location:Netherlands status: Published online

Purpose. Pathogenic variants (PVs) in BRCA2 and ATMgenes have been linked to an increased risk of various cancers. BRCA2and ATM are part of the homologous recombination pathway, but the tumor risk in patients with simultaneous PVs in both genes remains largely unknown. In this study we describe four patients from three families with multiple cancers who coinherited PVs in BRCA2 and ATM genes. Methods. PVs in the patients were identified using NGS sequencing of the DNA and were confirmed by Sanger sequencing. Results. The first family included a 67-year-old male with kidney, prostate, and pancreatic adenocarcinomas, and his daughter diagnosed with breast cancer at 29 years. In the second family, a 28-year-old female had breast cancer, while a male from the third family was diagnosed with prostate cancer at the age of 49, gastric cancer one year later and pancreatic cancer at 64. The three identified BRCA2 PVs were nonsense variants previously described as pathogenic, leading to a severely truncated or absent protein due to nonsense-mediated mRNA decay. Two of the ATM variants were previously reported as pathogenic, while the third one affects a conserved splice site. Conclusions. The early age of diagnosis and the development of multiple cancers in the reported patients indicate a very high risk of cancer in double heterozygous patients, carrying PVs in the BRCA2 and ATM genes.

The copy number and mRNA expression of STAT5b were assessed in samples from the TCGA repository of glioblastomas (GBM). The activation of this transcription factor was analyzed on tissue microarrays comprising 392 WHO 2016 GBM samples from our clinical practice. These data were correlated with patient survival using multivariable Cox analysis and, for a subset of 167 tumors, with signs of tumor invasiveness on the MRI. The effects of STAT5b knockdown by siRNA were assessed on the growth, therapeutic resistance, invasion and migration of GBM cell lines U87, U87‑EGFRVIII and LN18 and primary cultures GM2 and GM3. The activation, but not the copy number or the mRNA expression of nuclear transcription factor STAT5b expression correlated inversely with patient survival independently of IDH1R132H status, age, Karnofsky Performance Score, treatment and tumor volume. STAT5b inhibition neither altered the cell proliferation nor reduced the clonogenic proliferative potency of GBM cells, and did not sensitize them to the cytotoxic effect of ionizing radiation and temozolomide

A better understanding of transcriptional evolution of IDH-wild-type glioblastoma may be crucial for treatment optimization. Here, we perform RNA sequencing (RNA-seq) (n = 322 test, n = 245 validation) on paired primary-recurrent glioblastoma resections of patients treated with the current standard of care. Transcriptional subtypes form an interconnected continuum in a two-dimensional space. Recurrent tumors show preferential mesenchymal progression. Over time, hallmark glioblastoma genes are not significantly altered. Instead, tumor purity decreases over time and is accompanied by co-increases in neuron and oligodendrocyte marker genes and, independently, tumor-associated macrophages. A decrease is observed in endothelial marker genes. These composition changes are confirmed by single-cell RNA-seq and immunohistochemistry. An extracellular matrix-associated gene set increases at recurrence and bulk, single-cell RNA, and immunohistochemistry indicate it is expressed mainly by pericytes. This signature is associated with significantly worse survival at recurrence. Our data demonstrate that glioblastomas evolve mainly by microenvironment (re-)organization rather than molecular evolution of tumor cells.

The emergence of the SARS-CoV-2 Delta variant of concern (lineage B.1.617.2) in late 2020 resulted in a new wave of infections in many countries across the world, where it often became the dominant lineage in a relatively short amount of time. We here report on a novel genomic surveillance effort in Rwanda in the time period from June to September 2021, leading to 201 SARS-CoV-2 genomes being generated, the majority of which were identified as the Delta variant of concern. We show that in Rwanda, the Delta variant almost completely replaced the previously dominant A.23.1 and B.1.351 (Beta) lineages in a matter of weeks, and led to a tripling of the total number of COVID-19 infections and COVID-19-related fatalities over the course of only three months. We estimate that Delta in Rwanda had an average growth rate advantage of 0.034 (95% CI 0.025-0.045) per day over A.23.1, and of 0.022 (95% CI 0.012-0.032) over B.1.351. Phylogenetic analysis reveals the presence of at least seven local Delta transmission clusters, with two of these clusters occurring close to the border with the Democratic Republic of the Congo, and another cluster close to the border with Tanzania. A smaller Delta cluster of infections also appeared close to the border with Uganda, illustrating the importance of monitoring cross-border traffic to limit the spread between Rwanda and its neighboring countries. We discuss our findings against a background of increased vaccination efforts in Rwanda, and also discuss a number of breakthrough infections identified during our study. Concluding, our study has added an important collection of data to the available genomes for the Eastern Africa region, with the number of Delta infections close to the border with neighboring countries highlighting the need to further strengthen genomic surveillance in the region to obtain a better understanding of the impact of border crossings on lowering the epidemic curve in Rwanda.

ObjectiveThe link between BRCA1 and homologous recombination deficiency (HRD) in cancer has gained importance with the emergence of new targeted cancer treatments, while the available data on the role of the gene in colorectal cancer (CRC) remain contradictory. The aim of this case series was to elucidate the role of known pathogenic BRCA1 variants in the development of early-onset CRC.DesignPatients were evaluated using targeted next generation sequencing, exome sequencing and chromosomal microarray analysis of the paired germline and tumor samples. These results were used to calculate the HRD score and the frequency of mutational signatures in the tumors.ResultsThree patients with metastatic CRC were heterozygous for a previously known BRCA1 nonsense variant. All tumors showed remarkably high HRD scores, and the HRD-related signature 3 had the second highest contribution to the somatic pattern of variant accumulation in the samples (23% in 1 and 2, and 13% in sample 3).ConclusionsA BRCA1 germline pathogenic variant can be involved in CRC development through HRD. Thus, BRCA1 testing should be considered in young patients with a personal history of microsatellite stable CRC as this could further allow a personalized treatment approach.

Molecular algorithms may estimate the risk of recurrence and death for patients with endometrial cancer (EC) and may impact treatment decisions. To detect microsatellite instabilities (MSI) and p53 mutations, immunohistochemistry (IHC) and molecular techniques are used. To select the most appropriate method, and to have an accurate interpretation of their results, knowledge of the performance characteristics of these respective methods is essential. The objective of this study was to assess the diagnostic performance of IHC versus molecular techniques (gold standard). One hundred and thirty-two unselected EC patients were enrolled in this study. Agreement between the two diagnostic methods was assessed using Cohen’s kappa coefficient. Sensitivity, specificity, positive (PPV) and negative predictive values (NPV) of the IHC were calculated. For MSI status, the sensitivity, specificity, PPV and NPV were 89.3%, 87.3%, 78.1% and 94.1%, respectively. Cohen’s kappa coefficient was 0.74. For p53 status, the sensitivity, specificity, PPV, and NPV were 92.3%, 77.1%, 60.0% and 96.4%, respectively. Cohen’s kappa coefficient was 0.59. For MSI status, IHC presented a substantial agreement with the polymerase chain reaction (PCR) approach. For the p53 status, the moderate agreement observed between IHC and next generation sequencing (NGS) methods implies that they cannot be used interchangeably.

Objective Screening studies have established genetic risk profiles for diseases such as multiple endocrine neoplasia type 1 (MEN1) and pheochromocytoma–paraganglioma (PPGL). Founder effects play an important role in the regional/national epidemiology of endocrine cancers, particularly PPGL. Founder effects in the Netherlands have been described for various diseases, some of which established themselves in South Africa due to Dutch emigration. The role of Dutch founder effects in South Africa has not been explored in PPGL. Design We performed a single-center study in South Africa of the germline genetic causes of isolated/syndromic neuroendocrine tumors. Methods Next-generation panel, Sanger sequencing and multiplex ligand-dependent probe amplification for endocrine neoplasia risk genes. Results From a group of 13 patients, we identified 6 with PPGL, 4 with sporadic or familial isolated pituitary adenomas, and 3 with clinical MEN1; genetic variants were identified in 9/13 cases. We identified the Dutch founder exon 3 deletion in SDHB in two apparently unrelated individuals with distinct ethnic backgrounds that had metastatic PPGL. Asymptomatic carriers with this Dutch founder SDHBexon 3 deletion were also identified. Other PPGL patients had variants in SDHB, and SDHD and three MEN1variants were identified among MEN1 and young-onset pituitary adenoma patients. Conclusions This is the first identification of a Dutch founder effect for PPGL in South Africa. Awareness of the presence of this exon 3 SDHB deletion could promote targeted screening at a local level. Insights into PPGL genetics in South Africa could be achieved by studying existing patient databases for Dutch founder mutations in SDHx genes.

At the end of 2020, several new variants of SARS-CoV-2—designated variants of concern—were detected and quickly suspected to be associated with a higher transmissibility and possible escape of vaccine-induced immunity. In Belgium, this discovery has motivated the initiation of a more ambitious genomic surveillance program, which is drastically increasing the number of SARS-CoV-2 genomes to analyse for monitoring the circulation of viral lineages and variants of concern. In order to efficiently analyse the massive collection of genomic data that are the result of such increased sequencing efforts, streamlined analytical strategies are crucial. In this study, we illustrate how to efficiently map the spatio-temporal dispersal of target mutations at a regional level. As a proof of concept, we focus on the Belgian province of Liège that has been consistently sampled throughout 2020, but was also one of the main epicenters of the second European epidemic wave. Specifically, we employ a recently developed phylogeographic workflow to infer the regional dispersal history of viral lineages associated with three specific mutations on the spike protein (S98F, A222V and S477N) and to quantify their relative importance through time. Our analytical pipeline enables analysing large data sets and has the potential to be quickly applied and updated to track target mutations in space and time throughout the course of an epidemic., SCOPUS: ar.j, info:eu-repo/semantics/published

The integration of a viral genome into the host genome has a major impact on the trajectory of the infected cell. Integration location and variation within the associated viral genome can influence both clonal expansion and persistence of infected cells. Methods based on short-read sequencing can identify viral insertion sites, but the sequence of the viral genomes within remains unobserved. We develop PCIP-seq, a method that leverages long reads to identify insertion sites and sequence their associated viral genome. We apply the technique to exogenous retroviruses HTLV-1, BLV, and HIV-1, endogenous retroviruses, and human papillomavirus., SCOPUS: ar.j, info:eu-repo/semantics/published

Background: Glioblastoma is the most prevalent and severe type of malignant brain tumor in adults. Although the genetic make-up initiating glioblastoma is increasingly better understood, a better understanding in the mechanisms that drive its evolution, heterogeneity and therapy resistance may reveal new directions for therapy development. To get better insights into glioblastoma evolution, we analyzed and de-convoluted transcriptomes of primary and recurrent glioblastoma resections. Material and Methods: Matching primary and secondary resections from n=185 glioblastoma patients were collected as part of EORTC Study 1542. The study was extended with tumor pairs from n=51 patients from the international GLASS study. The datasets were subjected to differential and deconvolution analysis using in-house algorithms. Results: When mapping the tumor samples into a reduced Glioblastoma Intrinsic Transcriptional Subtype space, we visualized subtype traversal, indicating that the CL subtype most often switches. As we found no more transitions from MES to other subtypes than to be expected by chance, we concluded that MES is an end-state. On average, tumor cell percentages decreased from ~67% to ~50% mostly due to an increase in TAM/microglia. Differential expression analysis was performed with correction for tumor cell percentages. While expression of most known oncogenes did not change considerably over time, marker genes of TAM/microglia, neurons and oligodendrocytes were up-regulated whereas endothelial cell markers were down-regulated over time. Furthermore, a cluster of ~30 extracellular matrix-associated genes increase significantly over time. A signature representing the gene-set was significantly associated with poor survival; high signatures were in particular associated to survival in secondary resections (P = 6.613e-06, Kaplan-Meier estimator). This suggests that the increase of extracellular matrix expression fulfils an important role in glioblastoma evolution. Conclusion: Using a large cohort, we interrogated changes in the glioblastoma transcriptome over time and found that in particular the composition of the tumor and its environment changes. The tumor cell percentage drops, suggesting more invasion or recruitment of non-malignant cells or a combination of both. This change is independent of an increase in the prognostic increase in extracellular matrix expression. Citation Format: Youri Hoogstrate, Kaspar Draaisma, Santoesha A. Ghisai, Iris de Heer, Levi van Hijfte, Wouter Coppieters, Melissa Kerkhof, Astrid Weyerbrock, Marc Sanson, Ann Hoeben, Slávka Lukacova, Giuseppe Lombardi, Sieger Leenstra, Monique Hanse, Ruth Fleischeuer, Colin Watts, Joseph McAbee, Nicos Angelopoulos, Thierry Gorlia, Vassilis Golfinopoulos, Johan M. Kros, Vincent Bours, Martin J. van den Bent, Pierre A. Robe, Pim J. French. Transcriptional evolution of glioblastoma reveals changes in bulk composition, mesenchymal sub-type as end-state, and a prognostic association with increased extracellular matrix gene expression [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 6140.