Table 5.
Univariate and multivariate analysis for survival
|
Multivariate analysis P
|
|||||||
|---|---|---|---|---|---|---|---|
| Parameter | Univariate P* | Base model | Nonsignificant† | ERK&pERK | PKCα | pAKT | Combination |
| Age | ≤.001 | <.001 | — | .001 | <.001 | <.001 | <.001 |
| Performance status | <.001 | <.001 | — | <.001 | <.001 | .002 | <.001 |
| Favorable cytogenetics | ≤.001 | .025 | — | .029 | .024 | .026 | .03 |
| Unfavorable cytogenetics | ≤.001 | <.001 | — | <.001 | <.001 | <.001 | <.001 |
| ERK2, low/high | .017 | — | .15 | — | — | — | — |
| ERK2 & pERK2, 2 groups | .01 | — | — | .04 | — | — | — |
| PKCα, low/high | .009 | — | — | — | .03 | — | — |
| pAKT, low/high | .006 | — | — | — | — | .05 | — |
| PKCα-ERK2&pERK2-pAKT, 4 groups | .002 | — | — | — | — | — | .013 |
| AHD equal to 2 or higher | .019 | — | .24 | — | — | — | — |
| %C-Kit (CD117) + by flow | .025 | — | .61 | — | — | — | — |
| LDH | .005 | — | .25 | — | — | — | — |
| Albumin | .007 | — | .22 | — | — | — | — |
| BUN | .001 | — | .97 | — | — | — | — |
| CR | .002 | — | .22 | — | — | — | — |
| β2-microglobulin | <.001 | — | .07 | — | — | — | — |
| INR | .05 | — | .18 | — | — | — | — |
| SCT‡ | <.001 | — | .18 | — | — | — | — |
LDH indicates lactate dehydrogenase; BUN, blood urea nitrogen.
Variables that were analyzed but found to be nonsignificant at the P = .05 level in univariate analysis included chemotherapy treatment protocol, continuous or categorized white blood cell count, hemoglobin, platelet count, percentage of blood blasts, percentage of marrow blasts, percentage of CD34+ cells by flow cytometry, serum bilirubin, fibrinogen, fibrin split products, D-dimers, calcium, prothrombin time, activated partial thromboplastin time, sample source (blood versus marrow, P = .47), and diagnosis date, or sample date, the latter two of which were included as controls.
Variables listed in this column were shown to be significant by univariate analysis but were not significant when added back one at a time into the final model consisting of these variables: age, performance status, favorable cytogenetics, and unfavorable cytogenetics. The P shown is for the addition of that parameter alone to the aforementioned four parameters.
Stem-cell transplantation is not significant if the data set is limited to a population similar to those who underwent transplantation, that is, those younger than 67 years of age and surviving longer than 20 weeks. Since SCT is not a variable available for consideration at diagnosis, it was not included in the final model using variables available at the time of diagnosis. The multivariate results do not change significantly if SCT is included, with the combination terms having similar statistical significance, but the term for age dropping out of the final models.