This review summarizes reports of recurrent DNA sequence copy number losses in human neoplasms detected by comparative genomic hybridization. Recurrent losses that affect each of the chromosome arms in 73 tumor types are tabulated from 169 reports. The tables are available online at http:// www.amjpathol.org and http://www.helsinki.fi/ϳlgl _www/CMG.html. The genes relevant to the lost regions are discussed for each of the chromosomes. The review is supplemented also by a list of known and putative tumor suppressor genes and DNA repair genes (see Table 1 , online). Losses are found in all chromosome arms , but they seem to be relatively rare at 1q , 2p , 3q , 5p , 6p , 7p , 7q , 8q , 12p , and 20q. Losses and their minimal common overlapping areas that were present in a great proportion of the 73 tumor entities reported in Table 2 (see online) are (in descending order of frequency): 9p23-p24 (48%) , 13q21 (47%) , 6q16 (44%) , 6q26-q27 (44%) , 8p23 (37%), 18q22-q23 (37%) , 17p12-p13 (34%) , 1p36.1 (34%), 11q23 (33%) , 1p22 (32%) , 4q32-qter (31%) , 14q22-q23 (25%) , 10q23 (25%) , 10q25-qter (25%) ,15q21 (23%) , 16q22 (23%) , 5q21 (23%) , 3p12-p14 (22%), 22q12 (22%) , Xp21 (21%) , Xq21 (21%) , and 10p12 ( Knowledge of chromosomal deletions has significantly contributed to the detection of tumor suppressor genes, since the inactivation of one allele, according to the twohit hypothesis, often results from a deletion on the chromosomal level.

DNA copy number changes were studied by comparative genomic hybridization (CGH) on bone marrow samples obtained from 72 patients with childhood acute lymphoblastic leukemia (ALL) at diagnosis. The patients had been admitted to the Helsinki University Central Hospital (Finland) between 1982 and 1997. CGH showed DNA copy number changes in 45 patients (62.5%) with a mean of 4.6 aberrations per patient (range, 1 to 22). The results of CGH and chromosome banding analysis were generally concordant, but CGH facilitated specific karyotyping in 34 cases. DNA copy number gains were more frequent than losses (gains:losses, 6:1). Gains of DNA sequences affected almost exclusively whole chromosomes and were most commonly observed in chromosomes 21 (25%), 18 (22.2%), X (19.4%), 10 (19.4%) and 17 (19.4%). The most common partial gain was 1q31-q32 (8.3%). The most common gains of chromosomes 21, 18, X, 10, 17, 14, 4, 6 and 8 appeared concurrently. High-level amplifications of small chromosome regions were sporadic, detected only in two patients (2.8%). Chromosome 21 was involved in both cases. The most common losses were 9p22-pter (12.5%) and 12p13-pter (11.1%). No statistically significant association between the CGH findings and the diagnostic white blood cell count was observed.

Comparative genomic hybridization (CGH) analysis was performed on bone marrow specimens from 19 children with acute myeloid leukemia (AML) at diagnosis. The results of CGH were compared to those of conventional cytogenetic analysis. The most common CGH aberrations were gains of whole chromosomes 6 and 8, both of which appeared three times. Two losses were seen twice; losses of whole chromosomes 7 and X. The CGH findings were concordant with the results of conventional karyotyping. CGH did not add new information to the karyotypes. Since no high-level amplification was found among the samples and standard karyotyping was highly successful, we do not advocate routine use of CGH in the diagnostic evaluation of childhood AML.

Pediatric patients with CML may transiently demonstrate cells positive for the Philadelphia chromosome but not actively transcribing the bcr/abl-fusion message in their marrow during their posttransplant evaluation but remain in remission. Recurrence is highly likely in patients demonstrating positivity for both; these patients may be considered candidates for donor lymphocyte transfusion therapy.