Table 2.
Predictors of Progression to Clinically Significant CMV Infection by Cox Regression Analysis
| Variable | Univariate Analysis | Multivariate Analysis | ||
|---|---|---|---|---|
| Crude HR (95% CI) | P Value | Adjusted HR (95% CI) | P Value | |
| Age | 0.99 (0.96–1.01) | .32 | ||
| Sex | .12 | …a | ||
| Male | Reference | |||
| Female | 1.74 (0.86–3.54) | |||
| Race | .89 | |||
| White | Reference | |||
| African American | 0.67 (0.09–4.97) | |||
| Hispanic | 0.64 (0.15–2.70) | |||
| Other | 0.81 (0.31–2.13) | |||
| Type of cancer | .25 | |||
| Leukemia | Reference | |||
| Lymphoma | 0.32 (0.08–1.35) | |||
| Other | 0.66 (0.23–1.90) | |||
| Type of transplant | .003 | …a | ||
| Matched related donor | Reference | |||
| Matched unrelated donor | 4.75 (1.43–15.73) | |||
| Autologous | 24.83 (3.90–158.15) | |||
| Steroids useb | 5.85 (2.61–13.11) | <.0001 | 4.87 (2.12–11.19) | .002 |
| GVHD | 1.39 (0.65–2.96) | .39 | ||
| HCT donor status | .009 | …a | ||
| CMV− | Reference | |||
| CMV+ | 0.35 (0.16–0.77) | |||
| Time from HCT to enrollment, d | 0.99 (0.97–1.02) | .47 | ||
| CMV-CMIb | <.0001 | <.0001 | ||
| High | Reference | Reference | ||
| Low | 9.68 (3.49–26.80) | 9.02 (3.19–25.51) | ||
High CMV-CMI was defined as a pp65 of >100 per 250000 cells and/or an IE1 SPC of >50 per 250000 cells. Low CMV-CMI was defined as a pp65 SPC of ≤100 per 250000 cells and an IE1 of ≤50 SPC per 250000 cells. CMV load was excluded from the multivariate Cox regression analysis as it was highly related to the ELISPOT CMV assay results.
Abbreviations: CI, confidence interval; CMV, cytomegalovirus; CMV-CMI, cytomegalovirus cell mediated immunity; GVHD, graft versus host disease; HCT, hematopoietic cell transplant; HR, hazard ratio; SPC, spot count.
aVariable was entered into the initial multivariate Cox regression model based on the P value of its univariate analysis (≤.20) and later removed from the final Cox regression model through the backward elimination procedure.
bSteroid use and CMV-CMI level (low/high) were time-dependent variables in the Cox regression analysis as they changed over time during the study period.