Patient characteristics
| . | Discovery cohort . | Validation cohort . | ||
|---|---|---|---|---|
| n . | Risk for relapse (95% CI) . | n . | Risk for relapse (95% CI) . | |
| Age, y* | P = .95 | P = .23 | ||
| <40 | 24 | 21.5 (7.8-39.5) | 8 | 9.6 (0.2-28.9) |
| ≥40 | 19 | 24.8 (7.5-47.1) | 17 | 35.9 (14.5-57.0) |
| Diagnosis | P = .02 | P = .12 | ||
| CML/MPD | 9 | 0 | 3 | 0 |
| ALL | 4 | 0 | 3 | 33.3 (0.8-77.4) |
| AML/multidimensional scaling | 23 | 39.5 (19.0-59.5) | 17 | 29.4 (10.7-51.) |
| Lymphoma | 7 | 14.2 (0.7-46.5) | 2 | 0 |
| Remission status (based on minimal residual disease) at CBT | P = .09 | P = .16 | ||
| CR | 26 | 12.8 (3.2-29.3) | 14 | 11.1 (2.8-44.1) |
| Active disease | 17 | 35.9 (14.4-57.0) | 11 | 32.9 (9.0-54.2) |
| Disease risk index | P = .02 | P = .06 | ||
| Low + intermediate | 23 | 10.1 (3.7-27.6) | 9 | 14.2 (3.1-35.8) |
| High + very high | 19 | 39.5 (19.9-55.2) | 14 | 44.6 (18.3-62.1) |
| Conditioning regimen intensity | P = .03 | P = .08 | ||
| Myeloablative | 11 | 21.3 (5.2-41.1) | 7 | 10 (0.5-36.7) |
| Reduced intensity | 27 | 49.7 (32.8-65.6) | 14 | 50.6 (17.1-75.6) |
| Nonmyeloablative | 5 | 77.1 (34.2-93.1) | 4 | 82.3 (31.6-95.5) |
| CB-NKimfrequency† | P = .0003 | P = .001 | ||
| ≥11.8% | 32 | 72.7 (37.1-90.3) | 18 | 85.7 (33.4-97.8) |
| <11.8% | 11 | 4.3 (0.3-18.0) | 7 | 0 (—) |
| NK diversity index‡ | P = .05 | |||
| <805 | 10 | 60.0 (25.3-82.7) | ||
| ≥805 | 33 | 9.6 (2.4-22.8) | ||
| NK absolute count¶ | P = .85 | P = .14 | ||
| Low | 22 | 23.7 (8.6-43.0) | 13 | 9.1 (5.3-33.3) |
| High | 21 | 21.1 (6.5-41.1) | 12 | 48.4 (17.4-74.1) |
| . | Discovery cohort . | Validation cohort . | ||
|---|---|---|---|---|
| n . | Risk for relapse (95% CI) . | n . | Risk for relapse (95% CI) . | |
| Age, y* | P = .95 | P = .23 | ||
| <40 | 24 | 21.5 (7.8-39.5) | 8 | 9.6 (0.2-28.9) |
| ≥40 | 19 | 24.8 (7.5-47.1) | 17 | 35.9 (14.5-57.0) |
| Diagnosis | P = .02 | P = .12 | ||
| CML/MPD | 9 | 0 | 3 | 0 |
| ALL | 4 | 0 | 3 | 33.3 (0.8-77.4) |
| AML/multidimensional scaling | 23 | 39.5 (19.0-59.5) | 17 | 29.4 (10.7-51.) |
| Lymphoma | 7 | 14.2 (0.7-46.5) | 2 | 0 |
| Remission status (based on minimal residual disease) at CBT | P = .09 | P = .16 | ||
| CR | 26 | 12.8 (3.2-29.3) | 14 | 11.1 (2.8-44.1) |
| Active disease | 17 | 35.9 (14.4-57.0) | 11 | 32.9 (9.0-54.2) |
| Disease risk index | P = .02 | P = .06 | ||
| Low + intermediate | 23 | 10.1 (3.7-27.6) | 9 | 14.2 (3.1-35.8) |
| High + very high | 19 | 39.5 (19.9-55.2) | 14 | 44.6 (18.3-62.1) |
| Conditioning regimen intensity | P = .03 | P = .08 | ||
| Myeloablative | 11 | 21.3 (5.2-41.1) | 7 | 10 (0.5-36.7) |
| Reduced intensity | 27 | 49.7 (32.8-65.6) | 14 | 50.6 (17.1-75.6) |
| Nonmyeloablative | 5 | 77.1 (34.2-93.1) | 4 | 82.3 (31.6-95.5) |
| CB-NKimfrequency† | P = .0003 | P = .001 | ||
| ≥11.8% | 32 | 72.7 (37.1-90.3) | 18 | 85.7 (33.4-97.8) |
| <11.8% | 11 | 4.3 (0.3-18.0) | 7 | 0 (—) |
| NK diversity index‡ | P = .05 | |||
| <805 | 10 | 60.0 (25.3-82.7) | ||
| ≥805 | 33 | 9.6 (2.4-22.8) | ||
| NK absolute count¶ | P = .85 | P = .14 | ||
| Low | 22 | 23.7 (8.6-43.0) | 13 | 9.1 (5.3-33.3) |
| High | 21 | 21.1 (6.5-41.1) | 12 | 48.4 (17.4-74.1) |
Two-year cumulative incidence of relapse according to patient characteristics in the discovery and validation cohorts.
The median age was 38.8 years (range, 20.2-67.9 years) in the learning cohort and 53.6 years (range, 18.4-68.1 years) in the validation cohort.
The median values of CB-NKim for the discovery and validation cohorts were 5.0% (range, 0.3%-41.0%) and 9.1% (range, 0.4%-43.7%), respectively
ROC analysis identified the optimal diversity index cutoff that predicted for the risk for relapse to be 805. The median diversity index was 1317 (range, 421-5050)
The variable was categorized by using the median absolute NK count as a cutoff; namely, 17.3 K/μL (range, 0.2-53.2) and 9.8 K/μL (range, 1.3-62.3) for the discovery and validation samples, respectively. A more predictive cutoff could not be identified using ROC analysis.