Sensitivity to bias due to UGT1A9 c.98T>C
SNP
Prevalence of the variant UGT1A9 C.98T>C allele is
very low. None of the studies in Figures S3 and S4 identified a single
variant homozygous subject – only TC heterozygotes were found. Their
prevalence (estimated at 3.9% in Figure S4) suggests that one could
expect 2 TC subjects in the present sample of 54 matched patients, but
probability of observing 2 variant carriers in the present sample is
only 27.4%, probability of observing none is 11%, probability of
observing not more than one is 35.8% and probability of observing 2 or
more is 36.7%. Overall, TC patients were most likely only a few – if
any – in the present sample. There is no biologically plausible reason
to expect prevalence of UGT1A9 c.98T>C variant
carriers to differ between ABCG2 c.421C>A variant
carriers (“treated”) and wt subjects (“controls”). However, one
could assume a range of different imbalances in prevalence occurring by
chance. Table S2 illustrates two types of scenarios: in one, the
“starting” effect of the ABCG2 c.421C>A variant
allele that needs to be adjusted for bias is GMR=1.40 (as estimated in
the main analysis); in the other one, the “starting” effect of theABCG2 c.421 variant allele is GMR previously adjusted for bias
arising from not accounting for SLCO1A9 c.334T>G SNP
with a high imbalance in TT/TG prevalence between ABCG2 c.421variant carriers and wt controls (50% vs. 20%) and a marked effect of
the TT/TG genotype (ROM=1.25), i.e., GMR=1.31. The effect of the TCUGT1A9 c.98 genotype (associated with higher MPA
AUC0-12) is ROM=1.098, as estimated in Figure S3, or
GMR=1.50, as suggested by van Schaik [5] based on time-averaged AUCs
over a 12-month period in CsA or tacrolimus co-treated patients; and
there is a considerable imbalance between ABCG2 c.421 variant
carriers and wt controls in prevalence of UGT1A9
c.98T>C heterozygotes: a 3-fold difference as 6% vs. 2%,
or a 5-fold difference as 5% vs. 1% or 10% vs. 2%. As demonstrated
in Table S2, even in the “worst case scenario”, with the TC effect of
GMR=1.50 and a huge imbalance between treated and controls (10% vs.
2%), already adjusted ABCG2 c.421 variant allele effect would
still be 1.26, i.e., beyond the conventional limit of equivalent total
exposure.
Table S2 . Sensitivity analysis of the effect of ABCG2
c.421C>A variant allele on MPA AUCτ,ss to
account for hypothetical bias arising from not accounting for theUGT1A9 c.98T>C SNP. The effect of the TC (vs. TT)UGT1A9 c.98 genotype is estimated at ROM=1.098 (95%CI
0.548-2.198) (Figure S3) or as GMR=1.50 [5], and prevalence of the
TC genotype is estimated at 3.9% (Figure S4). Shown are bias-corrected
effects (GMRs) of the ABCG2 c.421C>A variant (vs.
wt) assuming considerable imbalance between variant carriers and wt
controls in prevalence of the UGT1A9 c.98 TC genotype, and
assuming TC effect as estimated (i.e., 1.10 or 1.50). The ABCG2
c.421 effect that is being bias-corrected is either GMR=1.40 (as
estimated in the main analysis) or GMR=1.31 – if adjusted for
hypothetical bias arising from not adjusting for SLCO1B3
c.334T>G genotype, under a large TT/TG imbalance betweenABCG2 variant carriers and wt controls, and the TT/TG effect
ROM=1.25 (see Table S1).