Figure S3 . Meta-analysis of studies reporting mean±SD MPA
AUC0-12 at steady-state in renal transplant recipients
treated with IR MMF and co-treated with CsA or macrolactams (tacrolimus
or sirolimus) and genotyped for UGT1A9*3 (c.98T>C)SNP, classified as variant carriers (heterozygous, TC) vs. wild type
homozygous subjects. The effect measure is ratio of means (ROM).
van Schaik 2009 [5] used reduced sampling schedule (at pre-dose, 0.5
and 2 hours post-dose) to project AUC0-12 at 3 and 10
days, and at 1, 3, 6 and 12 months postoperatively. They graphically
reported geometric mean AUCs (no measures of spread or precision) at
these time-points for 5 variant vs. 170 wt patients co-treated with CsA
and, separately, for 5 variant vs. 158 wt patients co-treated with
tacrolimus (Figure 3 in [5]). We graph-read (using digitizing
software) the geometric mean values at Day 10 (considering the timing of
the present measurements) and used p-values to recover common SD – this
is shown in Figure S3.
The number of variant allele carriers was very low in each individual
study and across studies by co-treatment and overall (Figure S3). van
Schaik [5] reported that based on mixed-modelling of repeatedly
assessed AUCs over time (at 6 time-points), among tacrolimus co-treated
subjects, 5 TC patients vs. 158 wt subjects had by around 50% higher
time-averaged AUCs (adjusted for age, sex, creatinine clearance,
delayed-graft function and MPA measurement technique); and the same was
reported for 5 TC subjects vs. 170 wt subjects co-treated with CsA
[5]. However, considering the AUCs on postoperative day 10, the
TC-wt difference in macrolactam co-treated subjects appeared even
greater, but there was apparently no TC-wt difference among CsA
co-treated patients (Figure S3). Kuypers [6] reported higher AUC in
3 TC vs. 92 wt subjects (Figure S3), while Picard [1] reported
comparable AUCs in 10 TC vs. 105 wt subjects co-treated with CsA (Figure
S3), and lower AUCs in 2 TC vs. 68 wt subjects co-treated with
tacrolimus (Figure S3). When pooled, this scarce data rather
consistently indicate lack of a relevant difference between TC (N=15)
and wt subjects (N=275) when CsA is co-treatment (Figure S3). When
marcolactams are co-treatment (Figure S3), individual study results are
heterogeneous indicating also no major difference between TC (N=10) and
wt subjects (N=318). In agreement, the overall pooled estimate (Figure
S3) does not signal a major difference between TC (N=25) and wt subjects
(N=593), but heterogeneity is huge, with 95% prediction intervals
indicating from 7 times lower to 7 times higher AUCs in TC vs. wild-type
subjects, i.e., approximately equal probability of TC being associated
with a considerably lower or considerably higher exposure to MPA.
Based on studies including exclusively or predominantly Caucasians of
European descent [1,2, 5-12] (pertinent for the present sample),
prevalence of the TT/TG SLCO1B3 c.334T>G genotype is
rather consistently estimated at 31.1% (95%CI 28.2-34.2; prediction
interval 24.9-38.1) (Figure S4), and prevalence of variantUGT1A9*3 (c.98T>C) allele carriage is consistently
estimated at 3.9% (95%CI 3.1-4.9, prediction interval 3.0-5.2) (Figure
S4).