Variables associated with the achievement of MMR: univariate and multivariate analyses
| . | n (%) . | HR . | (95% CI) . | P value . |
|---|---|---|---|---|
| Univariate analysis | ||||
| KIR | ||||
| KIR2DL5B | 31 (21%) | 0.423 | (0.262-0.682) | <.001 |
| KIR2DS3 | 44 (30%) | 0.547 | (0.368-0.811) | .003 |
| KIR2DL2 | 83 (56%) | 0.607 | (0.428-0.859) | .005 |
| KIR2DS2 | 82 (55%) | 0.71 | (0.501-1.01) | .055 |
| KIR3DS1 | 41 (28%) | 0.717 | (0.481-1.07) | .1 |
| KIR2DS5 | 51 (34%) | 0.758 | (0.519-1.11) | .15 |
| KIR2DS1 | 56 (38%) | 0.833 | (0.575-1.21) | .33 |
| KIR2DL5A | 44 (30%) | 0.836 | (0.562-1.25) | .38 |
| KIR2DL3 | 138 (93%) | 1.29 | (0.544-3.05) | .57 |
| KIR2DS4DEL* | 129 (87%) | 1.1 | (0.668-1.82) | .7 |
| KIR2DS4 | 60 (41%) | 1.02 | (0.716-1.44) | .93 |
| Sokal | ||||
| Low vs High | 0.648 | (0.446-0.942) | .023 | |
| Low vs Intermediate | 0.842 | (0.591-1.198) | .34 | |
| Intermediate vs High | 0.77 | (0.512-1.16) | .21 | |
| BCR-ABL1 at 3 months IS | ||||
| 1-10% vs <1% | 0.459 | (0.3019-0.699) | .003 | |
| >10% vs <1% | 0.164 | (0.0749-0.361) | <.001 | |
| >10% vs 1-10% | 0.358 | (0.164-0.782) | <.001 | |
| Multivariate analysis | ||||
| KIR | ||||
| KIR2DL5B | 0.52 | (0.284-0.951) | .034 | |
| KIR2DL2 | 0.62 | (0.3804-1.01) | .055 | |
| BCR-ABL1 at 3 months, IS | ||||
| 1-10% vs <1% | 0.164 | (0.2395-0.585) | <.001 | |
| >10% vs <1% | 0.13 | (0.0605-0.279) | <.001 | |
| >10% vs 1-10% | 0.347 | (0.164-0.735) | <.001 |
| . | n (%) . | HR . | (95% CI) . | P value . |
|---|---|---|---|---|
| Univariate analysis | ||||
| KIR | ||||
| KIR2DL5B | 31 (21%) | 0.423 | (0.262-0.682) | <.001 |
| KIR2DS3 | 44 (30%) | 0.547 | (0.368-0.811) | .003 |
| KIR2DL2 | 83 (56%) | 0.607 | (0.428-0.859) | .005 |
| KIR2DS2 | 82 (55%) | 0.71 | (0.501-1.01) | .055 |
| KIR3DS1 | 41 (28%) | 0.717 | (0.481-1.07) | .1 |
| KIR2DS5 | 51 (34%) | 0.758 | (0.519-1.11) | .15 |
| KIR2DS1 | 56 (38%) | 0.833 | (0.575-1.21) | .33 |
| KIR2DL5A | 44 (30%) | 0.836 | (0.562-1.25) | .38 |
| KIR2DL3 | 138 (93%) | 1.29 | (0.544-3.05) | .57 |
| KIR2DS4DEL* | 129 (87%) | 1.1 | (0.668-1.82) | .7 |
| KIR2DS4 | 60 (41%) | 1.02 | (0.716-1.44) | .93 |
| Sokal | ||||
| Low vs High | 0.648 | (0.446-0.942) | .023 | |
| Low vs Intermediate | 0.842 | (0.591-1.198) | .34 | |
| Intermediate vs High | 0.77 | (0.512-1.16) | .21 | |
| BCR-ABL1 at 3 months IS | ||||
| 1-10% vs <1% | 0.459 | (0.3019-0.699) | .003 | |
| >10% vs <1% | 0.164 | (0.0749-0.361) | <.001 | |
| >10% vs 1-10% | 0.358 | (0.164-0.782) | <.001 | |
| Multivariate analysis | ||||
| KIR | ||||
| KIR2DL5B | 0.52 | (0.284-0.951) | .034 | |
| KIR2DL2 | 0.62 | (0.3804-1.01) | .055 | |
| BCR-ABL1 at 3 months, IS | ||||
| 1-10% vs <1% | 0.164 | (0.2395-0.585) | <.001 | |
| >10% vs <1% | 0.13 | (0.0605-0.279) | <.001 | |
| >10% vs 1-10% | 0.347 | (0.164-0.735) | <.001 |
In all proportional hazard analyses involving KIRs, patients negative for the genotype are assigned a hazard ratio (HR) of 1.0. The genes KIR3DL3, KIR3DP1, KIR2DL4, and KIR3DL2 have prevalence of close to 100% in our cohort (similar to their known prevalence in the general population) and were excluded from further analysis. Univariate analyses revealed KIR2DL5B, KIR2DS3, KIR2DL2, and early molecular response (EMR; BCR-ABL1% at 3 months) to be significantly correlated with the achievement of MMR. Models including all combination of these variables were tested. Results from the optimum model selected using the Akaike information criteria is listed here and include only the input variables of KIR2DL2, KIR2DL5B, and EMR. Only the latter two were shown to be independently associated with molecular response. The inferior prognostic risk conferred by having BCR-ABL1 >10% at 3 months is particularly strong, even though 15 of 18 of the EMR failure patients had exposure to the more potent TKI nilotinib (2 transformed to BC and 1 patient withdrew prior to switching). KIR2DL5B and EMR were also associated with the achievement of MR4.5 in a multivariate analysis (KIR2DL5BPOS: HR, 0.50; P = .05; BCR-ABL1 >10% at 3 months: HR, 0.07; P = .005; 1-10% BCR-ABL1 at 3 months: HR, 0.36; P = .0002). Age and female sex did not have a statistically significant correlation with the achievement of MMR; nilotinib exposure was associated with inferior achievement of MMR, as is expected from the treatment schema of the TIDEL-II study (data not listed). (The multivariate model that included all 3 KIR genes with EMR is not listed here, as it does not optimally describe the data, and collinearity interferes with accurate reporting of hazard ratios).
KIR2DS4DEL refers to the 2DS4 allelic variant bearing a 22-bp deletion, also referred to as KIR1D.