TARGET Reveals the Genomic Characteristics of Multiple Childhood Cancers

April 1, 2018, by Jaime M. Guidry Auvil, Ph.D., Malcolm A. Smith, M.D., Ph.D., and Daniela S. Gerhard, Ph.D.

The Therapeutically Applicable Research to Generate Effective Treatments (TARGET), a large-scale pediatric genomic characterization initiative, continues to demonstrate the importance of integrative analyses using comprehensive data and metadata generated from high-quality patient tissues in both solid and hematologic cancers. March is National Kidney Cancer Awareness month, and a good time to reflect on the successes of TARGET in renal tumors, such as Wilms tumor (WT) and other childhood cancers. TARGET investigators published papers in 2017 integrating the genomic and clinical data of cases with high-risk subsets of WT and cases with acute myeloid leukemia (AML) and in February 2018 published an analysis across the spectrum of childhood cancers studied to date. These publications (and previous reported findings from TARGET) proved what many oncologists suspected: many pediatric cancers have distinctive molecular characteristics from adults, and therefore it will likely be important to develop specific treatments for these childhood cancers.

The TARGET initiative’s approach is to provide molecular characterization of each patient’s tumor genome, transcriptome, and epigenome through analysis of both primary and, when available, relapsed tumor tissues together with their matched normal tissue. The TARGET project teams (PTs) had access to cases for which both tumor and normal tissues were available along with the patient’s clinical and outcome data, predominantly from clinical or biological studies run through the Children’s Oncology Group (COG). The PTs integrated the molecular and clinical data to discover mechanisms of disease development and potential areas for improved therapeutic intervention strategies substantiating the critical need for large-scale research studies of similar design for the pediatric cancer community. Here we summarize some of the results recently published.

Wilms tumor (WT) – the most common kidney cancer in children – harbors genetic mutations across a number of genes that are important in two major cellular processes that occur early in kidney development: one pathway regulates miRNA biogenesis and another interferes with normal maturation of the kidney (induction) by regulating gene transcription. The WT PT studied patients who were defined as “high-risk”, including tumors with diffuse anaplastic (DA) histology and tumor with favorable histology (FH) that recurred predominantly within 5 years of initial treatment. A WT PT paper published August 2017, as well as the 2 previous manuscripts^1,2, suggest that targeting these pathways may provide viable therapeutic opportunities for high-risk Wilms tumors. Key findings from this paper include:

• The genes most commonly mutated in the TARGET WT cohort are TP53 (47.5% DA, 1.7% FH), CTNNB1 (13.5%), DROSHA (10.1%), and FAM123B (13.5%), most of which were already known to be associated with WT.
• Novel genes uncovered included certain miRNA processing, transcription, and renal development genes [e.g. DGCR8 (4.5%), XPO5 (1.5%), DICER1 (2.5%), SIX1 (3.8%), SIX2 (2.9%), and MLLT1 (3.7%)].
• Other mutated genes with novel association to WT included BCOR (2.6%), BCORL1 (3.8%), NONO (2%), MAX (1.7%), COL6A3 (3.2%), ASXL1 (1.7%), MAP3K4 (1.7%), and ARID1A (1.8%).
• The WT PT also reported that chromosome copy number changes were found in a number of discovery cases [e.g., recurrent 1q gain (56/117 patients; 47.9%), MYCN amplification (19/117 patients; 16.2%), LIN28B gain (24.8% in discovery cohort), and loss of MIRLET7A family members (5.1%-22%; up to 4x more prevalent in DA than FH)].
• Germline mutations were found in about 10% of the WT high risk cases studied as well (PALB2 (1.2%) and CHEK2 (1.2%) being novel).

Given the relatively high number of genes with candidate driver mutations, future treatment protocols targeting the common processes or pathways affected by the gene mutations may be more efficient than focusing on individual gene mutations.

Acute myeloid leukemia (AML) – a blood cancer arising in the bone marrow with 50% treatment failure rate – occurs in patients of all ages. The TARGET AML PT published in December 2017 the first large-scale study establishing the prevalence of and relationships among recurrent, somatic genetic and epigenetic alterations in pediatric AML, including how the frequency of these mutations changed as the patients’ age of AML onset increased. The AML PT observed several features common to pediatric and adult AML, including low overall rate of mutation compared to other cancers and overlap of some recurrently mutated genes. Fewer than 40 genes were mutated in more than 2% of cases.

Pediatric and young adult AML exhibit critically important molecular characteristics that are distinct in three age-related groups (<2yrs, 3-14yrs, and 15-39yrs). TARGET investigators identified novel gene fusions, many involving known partner NUP98 and focal deletions (MBNL1, ZEB2, and ELF1) that were more prevalent in young individuals as compared to adults. In addition, novel variants in GATA2, FLT3, and CBL, along with recurrent mutations in MYC, NRAS, KRAS, and WT1 appeared more frequently in pediatric AML. In contrast, targetable IDH mutations that are relatively common in adult AML are practically nonexistent in the childhood disease, reducing the potential role of IDH inhibitors for pediatric AML. Similarly, mutations in other genes commonly observed in adult AML such as DNMT3A and TP53 were nearly absent in pediatric cases.

The AML PT discovered that certain combinations of variants affect patient prognosis, for example FLT3-ITD in combination with a mutation in NPM1 confers a probability of improved survival. FLT3 ITD by itself, or even in combination with other mutations (which are important in AML on their own), are associated with highly aggressive disease. The TARGET PT further found that certain deletions, mutations, and hypermethylation of promoter DNAs cooperatively impacted key signaling pathways in growth, immunity, and alternate splicing that can lead to leukemogenesis. These results suggest that the development of future therapeutic strategies may benefit through an age-tailored, targeted approach to the treatment of pediatric AML.

In February 2018, the first trans-TARGET study³ of somatic alterations in 1,699 pediatric leukemia and solid tumors across six histotypes was published online. The analyses were performed on samples from young patients (most <20yrs and enrolled on COG trials). The manuscript was published back-to-back with a study from German investigators analyzing European childhood cancer cohorts. The studies were complementary inasmuch TARGET’s cohort was >50% hematologic cancers, while the European cohort included >50% brain cancers. The molecular characterization details were not identical, but the “big picture” conclusions agreed remarkably well. Here we summarize the TARGET results.

The trans-TARGET study found that 142 genes in the pediatric cancers studied had mutation frequencies high enough to declare that these are “driver” genes. Interestingly, only 45% matched potential cancer drivers found in adult tumors, providing additional support to the findings highlighted above that pediatric cancers can have different initiation and progression processes than adult cancers. The trans-TARGET study confirmed that:

• The median somatic mutation rates among all TARGET pediatric cohorts generally range from 0.17 per million bases (MB) to 0.7/MB and are substantially lower than those observed in common adult cancers (1-10/MB).
• The frequency of germline variants which are risks for tumor development are ~10%.

In addition, the trans-TARGET study found that:

• Somatic copy number alternations (sCNAs) and structural variations (SV) comprised the majority (62%) of events observed, specifically single nucleotide mutations or small indels were less frequent. This finding provided strong support to the initial design of the project, to utilize whole genome sequencing whenever possible, since whole exome sequencing would not have allowed detection of sCNAs and SVs mutations. In many patients these genomic alterations are the driver events.
• The genomes of 11% of TARGET patients revealed chromothripsis (i.e. massive rearrangements caused by a single catastrophic event).
• Analyses revealed a large number of low frequency drivers within and among disease cohorts.
• Driver gene alterations produced disruptions in pathways that may be targetable with existing treatments. Specifically, TARGET investigators found 21 biologic pathways disrupted by driver alterations, across cancers (i.e. cell cycle, epigenetic regulation) or histotype-specific (i.e. JAK-STAT, Wnt/ β-catenin, and NOTCH signaling).
• Of clinical significance, the genes mutated in shared or separate pathways were different among histotypes. Certain signaling pathways (RAS, JAK-STAT, and PI3K) show distinctive somatic alterations between solid tumors (primarily ALK, NF1, and PTEN) and leukemias (nearly all mutations in FLT3, PIK3CA, PIK3R1, and RAS genes).
• The somatic alterations with highest prevalence across certain disease groups occurred in CDKN2A (predominantly as deletions): mostly affecting T-ALL (78%), B-ALL (42%), and OS (11%).
• Over half of the pediatric driver genes observed across TARGET cohorts were specific to a single histotype (e.g. TAL1 for T-ALL and ALK for NBL).
• The trans-TARGET analyses outlined some known and novel, statistically significant co-occurrences (e.g., USP7, TAL1 in T-ALL; ETV6, IKZF1 in AML, and CREBBP, EP300 in B-ALL) or mutual exclusivities (e.g. MYCN, ATRX, or SHANK2 in NBL; PAX5, TP53 in B-ALL) among more than 300 gene-pairs.
• TARGET findings further indicate that subclonal mutations could be contributing to tumorigenesis in various childhood cancers, with nearly half of point mutations in leukemia and NBL driver genes showing low mutant allele frequencies (MAFs, <0.3).

The TARGET initiative and other large-scale genomics projects are transforming precision oncology for childhood cancers by identifying therapeutic strategies based on insights that can only be gleaned through high-quality, large-scale integrative data analyses. By creating a comprehensive molecular compendium of molecular alterations from large cohorts of cancer patients and by making these data available for investigators who will continually improve upon the knowledge base of these cancers, more effective classification and treatment strategies can be developed. Additional integrative manuscripts will be published for TARGET in the coming months, and the community is encouraged to use and follow-up on those important observations as they become available. For more information, including additional TARGET publications and data used for these analyses, please visit and explore the TARGET website and Data Matrix at the Office of Cancer Genomics.

References

1. Ooms AH, Gadd S, Gerhard DS, Smith MA, Guidry Auvil JM, Meerzaman D, Chen QR, Hsu CH, Yan C, Nguyen C, Hu Y, Ma Y, Zong Z, Mungall AJ, Moore RA, Marra MA, Huff V, Dome JS, Chi YY, Tian J, Geller JI, Mullighan CG, Ma J, Wheeler DA, Hampton OA, Walz AL, van den Heuvel-Eibrink MM, de Krijger RR, Ross N, Gastier-Foster JM, Perlman EJ. Significance of TP53 Mutation in Wilms Tumors with Diffuse Anaplasia: A Report from the Children's Oncology Group. Clinical Cancer Research. 2016 Nov 15;22(22):5582-5591. (PMID: 27702824)
2. Perlman EJ, Gadd S, Arold ST, Radhakrishnan A, Gerhard DS, Jennings L, Huff V, Guidry Auvil JM, Davidsen TM, Dome JS, Meerzaman D, Hsu CH, Nguyen C, Anderson J, Ma Y, Mungall AJ, Moore RA, Marra MA, Mullighan CG, Ma J, Wheeler DA, Hampton OA, Gastier-Foster JM, Ross N, Smith MA. MLLT1 YEATS domain mutations in clinically distinctive Favourable Histology Wilms tumours. Nature Communications. 2015 Dec 4;6:10013. (PMID: 26635203)
3. Ma X, Liu Y, Liu Y, Alexandrov LB, Edmonson MN, Gawad C, Zhou X, Li Y, Rusch MC, Easton J, Huether R, Gonzalez-Pena V, Wilkinson MR, Hermida LC, Davis S, Sioson E, Pounds S, Cao X, Ries RE, Wang Z, Chen X, Dong L, Diskin SJ, Smith MA, Guidry Auvil JM, Meltzer PS, Lau CC, Perlman EJ, Maris JM, Meshinchi S, Hunger SP, Gerhard DS, Zhang J. Pan-cancer genome and transcriptome analyses of 1,699 paediatric leukaemias and solid tumours. Nature. 2018 Mar 15;555(7696):371-376. (PMID: 29489755)