Page 49 - JCTR-11-2

Journal of Clinical and
Translational Research CMV secular trends and race

any diagnosed and treated infection leading to race on clinical outcomes (serostatus × transplant quarter
hospitalization. Opportunistic infections were also × race). All analyses were conducted using Statistical
sub-classified for this study. Analysis System (SAS) version 9.4 (SAS Institute, United
(x) Graft loss: Defined as a return to chronic dialysis, States [US]). p<0.05.
re-transplantation, or death. Death-censored graft loss
was also assessed for this study. 3. Results
During the 10-year study period, a total of 2,392 kidney
2.3. Data collection and covariates
transplants were performed at the study institution. After
Data collection included baseline donor/recipient excluding 79 pediatric patients and 53 patients with missing
demographics and transplant characteristics, validated data or second transplants, the final cohort consisted
from the transplant center-specific Standard Transplant of 2,261 patients. Among these patients, 1,287 patients
Analysis Research files. These included age, sex, race, (57%) were AA, with 185 (14.4%) having high-risk CMV
weight, history of diabetes, dialysis vintage, wait time, serostatus, and 974 patients (43%) were non-AA, with 25.4%
calculated panel reactive antibody (PRA) score at transplant, being high-risk. The Consolidated Standards of Reporting
human leukocyte antigen mismatches, cold ischemic time, Trials diagram is provided in Figure 1. Table 1 compares
delayed graft function, and pancreas transplant status. baseline characteristics between AA versus non-AA
Donor characteristics included age, sex, race, donor patients. The average age was 51 years for AAs and 53 years
type, and kidney donor risk index in deceased donors. for non-AAs, with females comprising 43.4% and 37.6%,
Longitudinal data were collected through a combination respectively. There was no significant difference in weight,
of electronic record data extractions from the institution’s body mass index, kidney or pancreas transplant history,
data warehouse and manual chart abstraction. Electronic or donor characteristics between the AA and non-AA
data extractions included medications, laboratory data, groups. History of diabetes mellitus, years on dialysis, and
and clinical events, such as hospitalizations, emergency waitlist status were significantly higher in AAs compared to
department visits, graft loss, and death, occurring during non-AAs. AAs also had higher PRA levels (p<0.001) and
the post-transplant period. Chart abstraction was used significantly lower rates of CMV D+/R− high-risk serostatus
to assess biopsy findings, causes of hospitalizations, type than the non-AA subjects (p<0.001). Delayed graft function
of CMV infection, confirmation of CMV disease (and rates were also significantly higher in AAs.
organ[s] affected), CMV treatments, and other relevant Table 2 shows unadjusted univariate analyses to compare
medications of interest for future analyses. AAs and non-AAs for CMV infections and other related
2.4. Statistical analysis outcomes. AAs had significantly higher rates of graft loss
(p=0.022), death-censored graft loss (p<0.001), antibody-
Data are presented as percentages for nominal variables, mediated rejections (p=0.003), acute cellular rejections
with univariate comparisons using Fisher’s exact test or (p=0.007), and hospitalizations for non-infection causes
Pearson’s Chi-squared test, as appropriate. For continuous (p=0.026). There were no statistically significant differences
variables with normal distribution, results are reported in BK viremia (p=0.10), BK nephropathy (p=0.205),
as means and standard deviations, with comparisons CMV infection rates (p=0.125), CMV disease (p=0.848),
performed using Student’s t-test for two independent resistance (p=0.844), or breakthrough (p=0.234) between
samples. For non-normally distributed variables, results AAs and non-AAs. The sequential Cox regression model
were reported using medians and interquartile ranges, assessing the impact of AA race, either unadjusted or
with univariate statistical comparison conducted using adjusted, on relevant clinical outcomes is illustrated
the Mann-Whitney U-test. Normal distribution was in Table 3. AA had a 68% higher risk of CMV infection
assessed using normality plots and the Shapiro-Wilk after adjustment for CMV serostatus (adjusted hazard
test. Multivariable analysis for secular trend assessments ratio [aHR] = 1.68, confidence interval [CI]: 1.30 – 2.17),
was conducted using generalized linear models (GLM) with a reduced risk after controlling for acute rejection
with a logit link for the binary outcome (CMV D+/R− (aHR = 1.21, CI: 0.94 – 1.56). It was significant in the
yes/no), with race as the primary independent variable fully adjusted model, whereby AA had a 41% higher risk
and transplant quarter as a covariate. The GLM accounted of CMV infection than non-AAs (aHR = 1.41, CI: 1.05 –
for time correlations within patients. Time-dependent Cox 1.86). The risk of developing late CMV infection, graft loss,
regression analysis was employed for multivariable survival and death-censored graft loss were significantly higher in
analyses involving all time-to-event outcomes. In these AAs versus non-AAs. In contrast, the risk of BK infection,
Cox models, a three-way interaction term was added to CMV disease, and breakthrough prophylaxis CMV
assess differences in temporal trends by CMV serostatus by infection were similar based on race (Table 4).

Volume 11 Issue 2 (2025) 43 doi: 10.36922/jctr.24.00067