Abstract
Previous studies have identified broad cytogenetic risk groups in AML, by comparing outcome of pts with different recurring abnormalities, using cytogenetically normal pts as the reference group. Pts with better survival were considered as “favorable”, and those with worse outcome as “unfavorable” risk. To identify prognostic cytogenetic groups for complete remission (CR) and overall survival (OS), without the selection of a reference group, we used classification trees and tree-structured survival analysis (TSSA). We analyzed the outcome of 600 AML pts ≥60 years (yrs), enrolled in the prospective CALGB cytogenetic study 8461 and treated on CALGB front-line AML protocols. Analysis was restricted to cytogenetic aberrations occuring in ≥5 pts. Once prognostic cytogenetic abnormalities for CR and OS were identified, multivariable models were constructed. Median age was 68 (range, 60–86) yrs, and 98% had de novo AML. The most common karyotypes were normal (46%), complex with ≥3 abnormalities (complex ≥3, 19%), and ≥5 abnormalities (complex ≥5, 14%). Core binding factor (CBF) abnormalities, ie, inv(16) and t(8;21), occurred in 5%. Overall, 49.5% of pts achieved CR with only 7% (95%CI: 5%–9%) alive at 5 yrs. Table 1 shows prognostic cytogenetic risk groups identified by classification trees and TSSA for CR and OS, respectively.
Table 1: Prognostic Cytogenetic Groups by Tree Analysis
CR . | OS . | ||
---|---|---|---|
Risk group . | CR rate . | Risk group . | 5-yr OS (95% CI) . |
*P<0.05 compared to CBF; CI, confidence interval | |||
Complex≥3 | 30%* | CBF | 20 (6–20)% |
Abnormal, non CBF | 46%* | <5 abnormalities without CBF or −7 | 8 (5–10)% |
Normal | 57% | Complex≥5 | 0% |
CBF | 75% | −7 | 0% |
CR . | OS . | ||
---|---|---|---|
Risk group . | CR rate . | Risk group . | 5-yr OS (95% CI) . |
*P<0.05 compared to CBF; CI, confidence interval | |||
Complex≥3 | 30%* | CBF | 20 (6–20)% |
Abnormal, non CBF | 46%* | <5 abnormalities without CBF or −7 | 8 (5–10)% |
Normal | 57% | Complex≥5 | 0% |
CBF | 75% | −7 | 0% |
Table 2 shows the multivariable analyses adjusting for other baseline clinical variables. Only lower % circulating blasts (P=0.0002) and cytogenetic risk groups (P<0.0001) predicted CR. Compared to CBF pts, the odds of CR were significantly lower for pts with complex ≥3 and abnormal, non-CBF karyotypes, whereas the odds of CR for cytogenetically normal pts was not significantly different from CBF pts. For OS, cytogenetic risk groups (P<0.0001), older age (P=0.03), and higher WBC at diagnosis (P=0.02) predicted shorter survival. Compared to CBF pts, the risk of death was significantly higher for those with complex ≥5 karyotype, −7, and <5 abnormalities without CBF or −7.
Table 2: Multivariable Analyses
CR . | OS . | ||
---|---|---|---|
Significant Variables . | OR (95% CI) . | Significant Variables . | HR (95% CI) . |
*CBF is the reference group; OR, odds ratio of achieving CR; HR, hazard ratio of death; CI, confidence interval | |||
% circulating blasts | 0.90 (0.84–0.95) | Age | 1.28 (1.09–1.50) |
Cytogenetics group* | WBC | 1.03 (1.01–1.04) | |
Complex≥3 | 0.12 (0.05–0.33) | Cytogenetics group* | |
Abnormal, non CBF | 0.26 (0.10–0.65) | Complex≥5 | 4.53 (2.80–7.32) |
Normal | 0.44 (0.18–1.09) | −7 | 2.57 (1.39–4.76) |
<5 abnormalities without CBF or −7 | 1.62 (1.07–2.46) |
CR . | OS . | ||
---|---|---|---|
Significant Variables . | OR (95% CI) . | Significant Variables . | HR (95% CI) . |
*CBF is the reference group; OR, odds ratio of achieving CR; HR, hazard ratio of death; CI, confidence interval | |||
% circulating blasts | 0.90 (0.84–0.95) | Age | 1.28 (1.09–1.50) |
Cytogenetics group* | WBC | 1.03 (1.01–1.04) | |
Complex≥3 | 0.12 (0.05–0.33) | Cytogenetics group* | |
Abnormal, non CBF | 0.26 (0.10–0.65) | Complex≥5 | 4.53 (2.80–7.32) |
Normal | 0.44 (0.18–1.09) | −7 | 2.57 (1.39–4.76) |
<5 abnormalities without CBF or −7 | 1.62 (1.07–2.46) |
We conclude that pre-treatment cytogenetics are predictive of CR and OS of older AML pts, particularly identifying pts with complex ≥5 karyotype and −7, who benefit minimally if at all from standard chemotherapy. Such pts may be better suited for investigational therapy or only supportive care.
Author notes
Corresponding author