Trial Information

A Factorial Randomized Controlled Trial of Correcting Anemia and Native Vitamin D Supplementation in Kidney Transplant Recipients

Sample size estimation:

The previous trial (the CAPRIT study) showed that 2.0 g/dL increase of hemoglobin (Hb)
reduced 69% of 2-year decline in estimated glomerular filtration rate (eGFR) (Choukroun G,
et al. J Am Soc Nephrol, 2012). Given that the annual eGFR decline in our patients with Hb
level <10.5 g/dL was 1.66 (SD, 2.47) mL/min per 1.73 m2, the investigators hypothesized that
the 2-year eGFR decline in the conservative anemia management group and the aggressive
anemia correction group should be 3.32 (SD, 4.94) and 1.03 (SD, 4.94) mL/min per 1.73 m2,
respectively. Assuming 20% of dropout or lost-to-follow, the planned sample size of 246
patients would yield a power of 90% for group comparison by using t-test with a type I error
of 5%.

Regarding cholecalciferol supplementation, 1,000 IU/day would increase serum
25-hydroxyvitamin D level by 11.8 ng/mL in patients with BMI <30, as suggested by the
previous trial (Gallagher JC, et al. Ann Intern Med, 2012). The investigators found in our
prospective cohort study that the 98.2% of Japanese kidney transplant recipients had BMI
<30, and that 10 ng/mL increase in 25-hydroxyvitamin D level was significantly associated
with 0.75 mL/min/1.73 m2 less decrease in annual eGFR change independent of potential
confounders (in submission). Therefore, the investigators expect 1.77 mL/min per 1.73 m2 in
eGFR would be preserved by 1,000 IU/day of cholecalciferol supplementation for 2 years.
Based on this assumption, this study size will provide a power of 70%.

Estimating kidney function:

In primary analyses, eGFR will be calculated by using the Japanese equation as in sample
size calculation (Matsuo S, et al. Am J Kidney Dis, 2009). However, this formula has not yet
been validated in kidney transplant recipients. Therefore, the investigators will use the
creatinine-based CKD-EPI equation with Japanese coefficient (Stevens LA, et al. Nephrol Dial
Transplant, 2010. Horio M, et al. Am J Kidney Dis, 2010) and an available formula if
validated in Japanese kidney transplant recipients at the time of analysis.

Statistical analyses:

For group comparison in a primary analysis, the investigators will use t-test or Wilcoxon
rank sum test according to the distribution of eGFR change. In the further analyses, to
analyze the time course of eGFR with respect to treatment assignment, changes in eGFR over
time will be analyzed with a linear mixed model for repeated measures with both fixed and
random intercept and slope. The multivariate model will contain time-varying eGFR as
dependent variable and treatment group as well as the number of measurements (time) as
independent variables. The study hypothesis will be tested by adding appropriate interaction
terms between the exposures and time. For secondary endpoints, the investigators will use
t-test, Wilcoxon rank sum test, or log-rank test for group comparison, and generalized
linear models or Cox proportional hazards models to estimate each effect of the
interventions, appropriately. The investigators will also adjust for baseline levels or past
history of each outcome. Other potential confounders, such as age, sex, time since
transplantation, blood pressure, urinary protein level, and diabetes, will be adjusted in
sensitivity analyses.

The interaction will be checked between anemia management and cholecalciferol
supplementation as well as between each intervention and baseline levels of urinary protein,
eGFR, Hb, 25-hydroxyvitamin D, the use of active vitamin D compounds, and the length of time
since transplantation. Additionally, stratified analyses will be conducted according to 0.2
g/g･creatinine of urinary protein and the date of transplantation (November 1999, the
release date of mycophenolate mofetil in Japan). However, the study size is not large enough
to statistically evaluate these interactions. The results from these analyses should be
interpreted with caution and regarded as exploratory and hypothesis generating.

Missing values:

Missing values will not be imputed in primary analyses. In sensitivity analyses, the
investigators will use multiple imputation method and last-observation-carried-forward
method.