Characterization of unusual iAMP21 B‐lymphoblastic leukemia (iAMP21‐ALL) from the Mayo Clinic and Children's Oncology Group

Abstract Acute lymphoblastic leukemia (B‐ALL) with intrachromosomal amplification of chromosome 21 (iAMP21‐ALL) represents a recurrent high‐risk cytogenetic abnormality and accurate identification is critical for appropriate clinical management. Identification of iAMP21‐ALL has historically relied on fluorescence in situ hybridization (FISH) using a RUNX1 probe. Current classification requires ≥ five copies of RUNX1 per cell and ≥ three additional copies of RUNX1 on a single abnormal iAMP21‐chromosome. We sought to evaluate the performance of the RUNX1 probe in the identification of iAMP21‐ALL. This study was a retrospective evaluation of iAMP21‐ALL in the Mayo Clinic and Children's Oncology Group cohorts. Of 207 cases of iAMP21‐ALL, 188 (91%) were classified as “typical” iAMP21‐ALL, while 19 (9%) cases were classified as “unusual” iAMP21‐ALL. The “unusual” iAMP21 cases did not meet the current definition of iAMP21 by FISH but were confirmed to have iAMP21 by chromosomal microarray. Half of the “unusual” iAMP21‐ALL cases had less than five RUNX1 signals, while the remainder had ≥ five RUNX1 signals with some located apart from the abnormal iAMP21‐chromosome. Nine percent of iAMP21‐ALL cases fail to meet the FISH definition of iAMP21‐ALL demonstrating that laboratories are at risk of misidentification of iAMP21‐ALL when relying only on the RUNX1 FISH probe. Incorporation of chromosomal microarray testing circumvents these risks.

FISH definition of iAMP21-ALL demonstrating that laboratories are at risk of misidentification of iAMP21-ALL when relying only on the RUNX1 FISH probe. Incorporation of chromosomal microarray testing circumvents these risks. tinct high-risk entity accounting for approximately 2% of B-ALL. It is associated with a median age of 9 years and a low white blood cell count (median 5 Â 10 9 /L). 3,4 Formation of the iAMP21-chromosome has been shown to arise from breakage-fusion-bridge cycles, resulting in chromothripsis of chromosome 21 with a variable region of copy number gain often, but not always, including the RUNX1 gene (at 21q22). 5,6 Deletion of the telomeric region of 21q, between genomic positions 44 and 47 Mb, has also been frequently observed. 5,[7][8][9] As the RUNX1 gene is often found in the highest region of chromosome 21 gain, fluorescence in situ hybridization (FISH) probes directed to the RUNX1 gene as part of the ETV6::RUNX1 fusion probe set is an efficient and cost-effective method to identify iAMP21-ALL within clinical cytogenetics laboratories. Currently, iAMP21-ALL is defined as greater than or equal to three extra copies of the RUNX1 gene on a single abnormal chromosome 21 (iAMP21-chromosome) and greater than or equal to five copies of the RUNX1 gene region per interphase cell. 3 However, accurate identification of iAMP21-ALL can be challenging as confirmation often requires FISH of abnormal metaphase cells, which are not always available, to distinguish polysomy 21 associated with favorable high hyperdiploidy from iAMP21-ALL.
Metaphase FISH also allows the determination of the chromosomal location of the additional RUNX1 signals, which are typically restricted to a single iAMP21-chromosome. However, rare, unusual cases of iAMP21-ALL have been previously described in which the extra RUNX1 signals are located on more than one abnormal chromosome. 10,11 While these unusual iAMP21-ALL cases did not fit the current FISH definition of iAMP21, chromosomal microarray analysis (CMA) confirmed the characteristic chromosome 21 copy number profile seen in iAMP21-ALL. As FISH is the most common, rapid, and cost-effective method used in the genomic characterization of B-ALL, we evaluated the performance of the RUNX1 probe in the identification of iAMP21-ALL and determined the frequency of unusual iAMP21-ALL in two large cohorts of B-ALL. Here, we describe 14 cases of B-ALL with genomic features of iAMP21-ALL by CMA that failed to meet the current definition of iAMP21-ALL using FISH.
We discuss an algorithmic approach to the genomic evaluation of B-ALL to reduce the misidentification and subsequent risk of misclassification of these unusual iAMP21-ALL cases.

| Patient selection
This study represented an institutional review board (IRB) approved retrospective evaluation of iAMP21-ALL from unique patients in two cohorts, Mayo Clinic and Children's Oncology Group (COG). See study design and participant selection ( Figure S1). For the Mayo Clinic cohort, we performed a search of all cases of pediatric B-ALL from patients (≤30 years of age) with evidence of clinical trial enrollment (n = 777) and all B-ALL from patients with evidence of iAMP21-ALL without clinical trial enrollment from January 2018 to December 2020. For the COG cohort, we performed a search of all patients with B-ALL that were enrolled on the COG APEC14B1 study (Project: Every Child: a registry, eligibility screening, biology, and outcome study) between August 2018 and June 2021.
Cases with genetic evidence of iAMP21-ALL and with CMA were further analyzed. Data were obtained from the COG registry and no additional experiments were performed on the COG cases. The methods below are specific for the Mayo Clinic cases.

| Chromosome analysis
White blood cells from the diagnostic bone marrow (BM) aspirate or lymph node specimen were cultured (24-and 48-h, unstimulated), harvested and G-banded slides were prepared using standard cytogenetic techniques.
Where possible 20 metaphases were analyzed and two karyograms prepared from each patient. 12 Chromosome analysis of BM specimens was performed in COG-approved local laboratories. Karyotypes were noted to represent unusual iAMP21-ALL at the time of central review.

| Mate pair sequencing
DNA was processed using the Illumina Nextera Mate Pair library preparation kit (Illumina, San Diego, CA) and sequenced on the Illumina HiSeq 2500 using 101-basepair reads and paired-end sequencing. Data were aligned to the reference genome (GRCh38) using BIMAv3 13 and abnormalities were identified and visualized using in-house developed bioinformatics pipeline, BMD-SV Pipeline. 14-16 Structural variants involving chromosome 21 were plotted using Circos. 17

| Statistical analysis
A Fisher's exact test was used to determine the significance between the number of males and females between the two iAMP21-ALL groups and a two-sided unpaired t-test was performed to determine the significance between the number of RUNX1 signals from the typical and unusual iAMP21-ALL cases (https://www.graphpad.com/quickcalcs/).

| Frequency of unusual iAMP21-ALL
To determine the frequency of unusual iAMP21-ALL, we performed a retrospective evaluation of the Mayo Clinic and the COG databases for cases classified as iAMP21-ALL. In total, 207 non-overlapping cases of iAMP21-ALL were identified using the RUNX1 FISH probe.
Using the current definition of iAMP21-ALL (greater than or equal to three extra copies of RUNX1 on a single abnormal chromosome 21 and greater than or equal to five copies of the RUNX1 gene region per cell), 188 (91%) cases met this definition and were classified as "typical" iAMP21-ALL, while 19 (9%) cases failed to meet this definition and were classified as "unusual" iAMP21-ALL ( Figure 1A, Figure S1).
Of these 19 unusual cases, five cases were removed from our cohort due to the lack of residual specimens for further characterization.

| Unusual iAMP21-ALL does not meet the current definition of iAMP21-ALL by FISH
The total number of RUNX1 signals could be accurately quantified by interphase FISH in 37 (93%) iAMP21-ALL cases, with a median number of five RUNX1 signals (range 3-10) per cell. However, in three (7%) cases, the total number of RUNX1 signals could not be accurately quantified due to their large number of RUNX1 signals (>10) per cell; thus, these cases were indicated simply as "RUNX1 amplification." The average maximum number and range of RUNX1 signals by interphase FISH were significantly higher in cases with typical iAMP21-ALL compared to unusual iAMP21-ALL (typical: average 6.2, range 5-10, unusual: median 4.8, range 3-8) (p < 0.0034) ( Figure 1B, C).
All 26 typical iAMP21-ALL cases met the current definition of iAMP21-ALL, which required a total of five or more RUNX1 signals per cell by interphase FISH or three or more extra RUNX1 signals on the iAMP21-chromosome by metaphase FISH (Table 1, data not shown). In contrast, none of the 14 unusual iAMP21-ALL cases met the current definition of iAMP21-ALL (Tables 1 and 2). Of these 14 unusual iAMP21-ALL cases, 7 (50%) cases (patients 2, 3, 7, 9, 10, 11, and 14) had fewer than 5 total RUNX1 signals per cell by interphase FISH ("Category 1") ( Figure 1D Table 2). The median number of fluctuating copy state changes throughout chromosome 21 was 7. The highest number of fluctuating copy state changes was found in four cases (3, 5, 6, and 7). Terminal deletions of distal 21q were also identified in 71% (10/14) of unusual cases and interstitial deletions of variable size of 21q were identified in 43% (6/14) of unusual cases ( Figure 2A, Table 2). RUNX1 was not always located within the region of highest gain. In patient 3, interphase and metaphase FISH identified three copies of RUNX1, while CMA confirmed iAMP21-ALL without amplification of RUNX1 ( Figure 2A). Metaphase FISH targeting 21q22.13 (CTD-2226O4) and 21q22.3 (CTD-2119D4) confirmed iAMP21-ALL with deletion of distal 21q ( Figure 1E).
To further characterize the genomic complexity of the unusual iAMP21-ALL, we performed mate-pair sequencing (MPseq) on seven cases (5 unusual iAMP21-ALL and 2 typical iAMP21-ALL), focusing on those cases with evidence of a RUNX1 signal observed on a chromosome other than chromosome 21 ( Figure 2B). Sequencing data were analyzed for the detection of breakpoint junctions, which is defined as the location of two novel chromosomal breakpoints now joined, and graphically illustrated as previously described. 15,18 In contrast to 2 typical iAMP21-ALL cases, which had 18 and 21 junctions within   (Figure 2A, Table 2). Of the 5 unusual iAMP21-ALL cases sequenced, rearrangements between chromosome 21 and additional chromosomes including chromosomes 5,8,13,19,20,22, and X were identified ( Figure 2, Table 2). In addition, 3 of 5 unusual cases had a junction involving the ERG gene at 21q22.2, another gene commonly found within the region of amplification in iAMP21-ALL. If number of RUNX1 signals observed by interphase FISH is indicated as "amplification" that means that there were greater than 10 signals per interphase cell. b Confirmed with pathology report that this patient has B-ALL.  whether this rearrangement produced aberrant CRLF2 gene expression is unknown. Three cases had abnormalities involving chromosome 7, four cases had a deletion of ETV6 and three cases had loss of RB1. Although gain of RUNX1 signals has historically been used to define iAMP21-ALL, the variable common region of gain of chromosome 21 encompasses over 40 genes including ERG 9 and amplification of ERG has been previously described in cases of iAMP21-ALL. [23][24][25] ERG is a proto-oncogenic transcription factor that is involved in cell cycle development and regulation. As RUNX1 does not appear to be the oncogenic driver of iAMP21-ALL, 7 additional studies are warranted to evaluate whether ERG is a potential driver of iAMP21-ALL.

| DISCUSSION
In all cases in our study, ERG was within the region of chromosome gain, with a copy number ranging from 3 to 8.
To our knowledge, this is the first study to report genomic characterization of iAMP21-ALL using MPseq. MPseq utilizes a specialized library preparation followed by whole-genome sequencing to identify to chromosome X, conventional cytogenetic analysis identified a der(X)t (X;21)(p22.1;q22), with one RUNX1 signal identified on chromosome X. In this case, all three testing strategies were consistent. Whether the mechanism causing unusual iAMP21-ALL, including those cases with rearrangements involving other chromosomes, also involves breakagefusion-bridge cycles is also unknown. 5,8,9,28,29

| CONCLUSION
In summary, we demonstrate that FISH is insufficient for accurate identification of all iAMP21-ALL cases. We confirm the value of CMA for validation of ambiguous cases, particularly in those lacking a recurrent primary abnormality or those having monosomy 21 (or with a marker or ring chromosome with RUNX1 signals). Our study demonstrates that misclassification of iAMP21-ALL may occur because RUNX1 may not be located within the highest region of chromosome 21 gain and/or the additional RUNX1 signals are present on chromosomes other than the abnormal chromosome 21. We have proposed an algorithmic approach to aid in the accurate identification of iAMP21-ALL utilizing interphase and metaphase B-ALL FISH and CMA to help guide laboratory geneticists in the identification of unusual iAMP21-ALL ( Figure 3). This study is limited by its retrospective nature, small sample size of the unusual iAMP21-ALL subtype, and absence of patient outcome data. In addition, this study reports a total percent of unusual iAMP21-ALL cases that may be misclassified; however, this is the combined experience of a single laboratory (Mayo Clinic) and COG, which represents multiple laboratories. Thus, the individual incidences 7 of 33 or 21% for Mayo Clinic and 12 of 174 or 7% for COG differ and this observation may be explained by the variability in laboratory practices and in the sample sizes of the two cohorts (33 vs. 174). Future studies, including larger cohort sizes, are necessary to identify more precisely the incidence of iAMP21-ALL, including unusual cases, and to evaluate whether more cases without a primary genetic abnormality may harbor yet unidentified cases of unusual iAMP21-ALL. Further studies are also needed to evaluate the clinical significance of unusual iAMP21-ALL in comparison to typical iAMP21-ALL. Incorporation of CMA or utilization of alternative probe sets on chromosome 21 should be considered in cases with an unknown primary abnormality to evaluate the presence of iAMP21-ALL for accurate risk stratification.