Clinical characteristics of cytomegalovirus-positive pediatric acute lymphoblastic leukemia at diagnosis.

Abstract

Infections and antigenic exposures during childhood are associated with pediatric acute lymphoblastic leukemia (ALL) and are thought to lead to immune dysregulation stimulating pre-leukemic clones to expand and progress to overt leukemia.1 Emerging epidemiologic and laboratory evidence suggests cytomegalovirus (CMV) may contribute to the development of childhood ALL,2, 3 inspiring further investigative efforts and for the first time, identifying a specific target for ALL prevention. In this study, we aimed to better elucidate the role of CMV in ALL etiology by screening diagnostic leukemia bone marrow samples for CMV DNA using a highly quantitative droplet digital PCR (ddPCR) assay. We identified differences in demographic features and leukemia subtypes between CMV-positive and CMV-negative cases, supporting the hypothesis that CMV plays a role in ALL development. Diagnostic leukemia bone marrow samples were obtained from the California Childhood Leukemia Study (CCLS), which included children less than 15 years old with newly diagnosed leukemia. The QIAamp DNA Blood Mini Kit (Qiagen) was used to isolate DNA from 5 mL bone marrow aspirate samples acquired from 1078 patients. A ddPCR assay was used to screen samples for CMV DNA targeting a sequence recurrently identified in samples that had previously undergone whole genome sequencing. To normalize the CMV-positive droplet count to the amount of DNA in the reaction, a second ddPCR reaction was run on all samples to detect single copy human DNA target. A ratio of CMV to human haploid positive droplets (CMV-ratio) was calculated as a measure of the level of CMV-positivity. Additional methods are included in Supporting information. Both ALL and acute myeloid leukemia (AML) cases were included in the analysis with AML serving as a control as it represents a similarly immunocompromised state as ALL cases, but less frequently exhibits CMV DNA.2 Samples were classified as CMV-positive if the CMV-ratio was greater than 0 and CMV-negative if equal to 0. CMV-ratio was also categorized into quintiles or tertiles based on the distribution in the overall cohort with a CMV-ratio of 0 as the referent for analyses of associations with CMV viral DNA load. Clinical and demographic features were compared between the CMV-positive and CMV-negative cases including leukemia phenotype, age at diagnosis, sex, race, and ethnicity. Using existing data for ALL cases, we assessed the distribution of somatic gene deletions (n = 702), RAS mutations (n = 469), and FLT3 alterations (n = 206) between CMV-positive and CMV-negative groups. Differential gene expression was performed using available Affymetrix Array gene expression data (n = 61) to compare CMV-positive ALL cases with high viral load to CMV-negative cases. SNP array data and ALL polygenic risk scores (PRS) were available for 435 ALL cases. The PRS and risk alleles for individual SNPs were compared between CMV-positive and CMV-negative groups. Genetic ancestry (Latino and non-Latino) determined from the SNP data was included in this analysis. Wilcoxon rank sum (continuous variables) and chi-square tests (categorical variables) were used for univariate analyses and logistic regression for multivariable analysis. All statistical tests were two-sided and results were considered statistically significant for p < .05. Differential gene expression analysis was performed using logistic regression for individual genes and gene ontology analysis was performed using the software IPA (QIAGEN Inc., https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis). A total of 743 ALL samples and 125 AML samples were included in the analysis. Of these, 49% (n = 424) were CMV-positive and 51% (n = 444) were CMV-negative. Age, sex, race, and ethnicity were compared by CMV status in the overall group and separately among ALL and AML cases (Table S1). Among ALL cases, CMV-positive children were older than their CMV-negative counterparts (median age at diagnosis 4.9 and 4.6 years respectively, p = .04). Latinos were 1.44 times as likely to be CMV-positive compared with non-Latinos in the overall cohort (95% CI: 1.06–1.97, p = .02). Among ALL cases, there was some suggestion that CMV-positivity might be more common in Latinos although results were not statistically significant (OR: 1.28, 95% CI: 0.92–1.79, p = .13). ALL cases were 2.5 times more likely to be strongly CMV-positive (highest quintile) than CMV-negative when compared with AML cases (95% CI: 1.00–5.47, p = .039), but is no longer statistically significant after adjusting for ethnicity (OR: 1.93, 95% CI: 0.44–1.29, p = .179, Table S4). This differs from previous findings, which may be due to differences in methods, most notably that Francis et al utilized RNA rather than DNA from bone marrow for CMV detection.2 Among ALL cases, B-cell ALL was more likely to be strongly CMV-positive (highest tertile) than CMV-negative when compared with T-cell ALL (OR: 2.93, 95% CI: 1.01–8.52, p = .048, Figure 1A). Upon comparison of ETV6-RUNX1 to high hyperdiploid (HHD) cases (the two most common B-ALL subtypes), HHD cases were 2.71 times more likely to be strongly CMV-positive (highest tertile) than CMV-negative after adjusting for age and ethnicity (95% CI: 1.34–4.73, p = .004, Figure 1B). The association of high hyperdiploidy with CMV status was unexpected, particularly with the known potentially virally-associated (APOBEC) DNA mutation signatures previously noted in ETV6-RUNX1 cases.4 Though both the HHD subtype and CMV are more common in Latinos,5 our analysis shows that the association between CMV and HHD ALL is not driven by ethnicity. These novel associations between CMV-positivity at the time of leukemia diagnosis and B-cell ALL, specifically the HHD subtype suggest that CMV could play a role in the development of particular subtypes of ALL. Using existing data for ALL cases, we assessed the distribution of somatic copy-number loss in EBF1, IKZF1, CDKN2A, PAX5, ETV6, BTG1, RB1, and the pseudoautosomal region 1 (PAR1) overlapping CRLF2 (n = 702), RAS mutations (n = 469), and FLT3 alterations (n = 206) between CMV-positive and CMV-negative groups. Non-HHD ALL cases harboring EBF1 deletions were more likely to be CMV-positive (second and third tertiles) than CMV-negative compared with cases without EBF1 deletions (OR: 6.10; 95% CI: 1.09–34.06, p = .04 for CMV second tertile; OR: 5.54; 95% CI: 1.13–27.18, p = .03 for CMV highest tertile) though this association would not remain significant after adjusting for multiple comparisons. No statistically significant associations between CMV-ratio and deletions or alterations in the other genes listed above or total number of deletions were identified. Differential gene expression analysis revealed 830 genes to be significantly differentially expressed between the highest quintile of CMV-positive cases (n = 6) and CMV-negative cases (n = 29), and gene ontology analysis revealed upregulation of processes involved in viral infection and replication (Figure S1). Specifically, cytokine signaling pathways including IL-1, IL-7, IL-8, and B-cell receptor signaling were upregulated in CMV-positive cases, while Th1 and the pathway facilitating crosstalk between dendritic cells and natural killer cells were downregulated a pattern consistent with an acute CMV infection (Table S2). IL-7 is crucial for lymphopoiesis and its upregulation is described in both B-ALL and T-ALL,6 raising the possibility that CMV could contribute to ALL development through this pathway. CMV induces B-cell proliferation in acute infections, and in latent CMV infection, individuals have been shown to have persistent B-cell mediated immunologic abnormalities. In conjunction with the CMV and B-ALL association and upregulation of B-cell receptor pathways that we identified, this provides an interesting potential link between CMV and the development of pediatric ALL. Further, the gene expression patterns noted here suggest that CMV may contribute to the leukemic phenotype and, by proxy, a potential dependence of HHD leukemias on such gene expression patterns. Additional investigation of these novel findings may clarify the viral, immune, and leukemic phenotype associations. Among B-ALL cases, with increasing PRS, cases were 1.72 times more likely to be in the highest CMV quintile than to be CMV-negative after adjusting for subtype (95% CI: 1.003–2.96, p = .049, Table S3). When stratified by ethnicity, a similar association was seen among non-Latino cases (OR: 2.34, 95% CI: 1.16–4.73, p = .017) but was not demonstrated in Latino cases (OR: 1.23, 95% CI: 0.69–2.19, p = .484, Table S3). Furthermore, we assessed individual SNPs and found that those with a risk allele at the SNP 10:63721176:C:T in ARID5B were 2.0 times more likely to be CMV-positive than those with no risk allele after adjusting for subtype (95% CI: 1.06–3.88, p = .037). Stratified by ethnicity, non-Latinos with at least one risk allele were 3.12 times more likely to be CMV-positive than those with no risk allele (95% CI: 1.15–9.12, p = .029) after adjusting for subtype (Figure 1C). Furthermore, those with at least one risk allele in ARID5B were 10.9 times more likely to be in the highest CMV tertile compared with those without an ARID5B risk allele (95% CI: 1.29–91.86, p = .028). We propose a possible mechanism to explain this association; in children with a genetic predisposition, as indicated by higher PRS, an acute CMV infection acts to promote leukemic proliferation potentially through the stimulation of B-cells by CMV gene products and its effect on the host immune system. The presence of viral particles previously identified in newly diagnosed ALL patients2 and differences in gene expression further supports the idea that an active CMV infection may be present at diagnosis and could thereby be driving ALL development. In conclusion, our results support the hypothesis that CMV plays an enhanced role in ALL development in those with a genetic predisposition to ALL. The patterns of gene expression in our analysis, though performed on a small subset of patients, suggest that active CMV infection could contribute to or even drive leukemia development. These intriguing results require validation and warrant continued investigation of the role of CMV in pediatric ALL development and its potential as a therapeutic target. This work was supported by 5T32CA009659-25 and grants from the U.S. National Institute of Environmental Health Sciences (2R01ES009137-11A1, P50ES018172/EPA RD83615901 to Catherine Metayer) and the Impact Grant via UCSF Cancer Center (Stephen S. Francis and Joseph L. Wiemels). The authors acknowledge additional funding from USC Center for Genetic Epidemiology, USC Cancer Center CCSG Development Funds, USC Cancer Center Norris Chair Funds, and the Shirley McKernan Courage Foundation (CHLA). The authors thank the families that participated in the CCLS. Without their time and effort, none of our studies would have been possible. For recruitment of subjects enrolled in the California Childhood Leukemia Study (CCLS), the authors gratefully acknowledge the clinical investigators at the following collaborating hospitals: University of California Davis Medical Center (Dr. Jonathan Ducore), University of California San Francisco (Drs. Mignon Loh and Katherine Matthay), Children's Hospital of Central California (Dr. Vonda Crouse), Lucile Packard Children's Hospital (Dr. Gary Dahl), Children's Hospital Oakland (Drs. James Feusner and Carla Golden), Kaiser Permanente Roseville (formerly Sacramento) (Drs. Kent Jolly and Vincent Kiley), Kaiser Permanente Santa Clara (Drs. Carolyn Russo, Alan Wong, and Denah Taggart), Kaiser Permanente San Francisco (Dr. Kenneth Leung), Kaiser Permanente Oakland (Drs. Daniel Kronish and Stacy Month), California Pacific Medical Center (Dr. Louise Lo), Cedars-Sinai Medical Center (Dr. Fataneh Majlessipour), Children's Hospital Los Angeles (Dr. Cecilia Fu), Children's Hospital Orange County (Dr. Leonard Sender), Kaiser Permanente Los Angeles (Dr. Robert Cooper), Miller Children's Hospital Long Beach (Dr. Amanda Termuhlen), University of California, San Diego Rady Children's Hospital (Dr. William Roberts), and University of California, Los Angeles Mattel Children's Hospital (Dr. Theodore Moore). The authors have no conflicts of interest to disclose. The raw de-identified data that support the findings of this study are available from the corresponding author upon request. The raw de-identified data that support the findings of this study are available from the corresponding author upon request. Table S1. Demographic and clinical characteristics by CMV status for 868 pediatric leukemia cases from the California Childhood Leukemia Study (CCLS) conducted from 1995–2015 Table S2. Enrichment of biological pathways of differentially expressed genes between highest acute lymphoblastic leukemia (ALL) cases harboring the highest CMV content (N = 6) compared to CMV-negative ALL cases (N = 29) Table S3. Association of ALL Polygenic Risk Score and CMV Status, California Childhood Leukemia Study 1995–2015 Table S4. Comparison of ALL and AML, B-cell and T-cell ALL, and ETV6-RUNX1 and High Hyperdiploid (HHD) ALL Cases, California Childhood Leukemia Study 1995–2015 Figure S1. Summary of pathways and disease processes enriched within differentially expressed genes between highest CMV ratio group (N = 6) and CMV negative (N = 29) acute lymphoblastic leukemia (ALL) cases Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.