Page 64 - JCBP-3-3

Journal of Clinical and
Basic Psychosomatics Depression and chronic pancreatitis: MR study

Based on the InSIDE hypothesis, MR-Egger regression Our MR analysis using the IVW method showed a
provided a consistent causal effect estimate, and its intercept positive association between depression with CP. The odds
represented the effect estimate of horizontal pleiotropy. ratio (OR) for CP per one-standard deviation increase in
27
MR-Egger regression can still yield unbiased estimates depression was 1.39 (95% confidence interval [CI]: 1.03 –
when horizontal pleiotropy of IVs is present. Simple 1.86; P = 0.03) (Figure 3). This result was consistent with
28
mode and weighted mode were used as complementary the weighted median sensitivity analysis, which reported
analyses. We tested multiple IVs for horizontal pleiotropy an OR of 1.56 (95% CI: 1.04 – 2.35; P = 0.03) (Figure 3).
using MR-Egger to assess potential violations of the MR Further validation using the MR-Egger test revealed no
hypothesis. 27,28 The heterogeneity between causal estimates evidence of horizontal pleiotropy (P = 0.89). The IVW
for each SNP was evaluated using the Cochran Q test. approach did not exhibit heterogeneity according to the
15
We also tested the sensitivity of causal inferences to any Cochran Q statistic (Q = 40.81, P = 0.44). The statistical
individual genetic variant through leave-one-out analysis. power of the MR analysis for depression on CP exceeded
15
The statistical power for each MR analysis was calculated 0.8, suggesting a relatively low probability of type II errors.
using an online MR power calculator for binary outcomes In the MR analysis, we did not find evidence of a causal
(https://shiny.cnsgenomics.com/mRnd/). 29 effect of CP on depression (IVW: OR: 1.004; 95% CI: 0.99 –
1.02; P = 0.63) (Tables S7 and S8, Figures S1 and S2).
2.2.2. Multivariable MR
The screening process for IVs followed a similar approach 3.1. Multivariable MR analysis
as described above. IVs needed to be genome-wide In multivariable MR analyses, the association between
significant in at least one exposure and presented in depression and CP remained significant after adjusting
all exposures and outcomes, as shown in Figure 2. The for five other exposures (triglycerides, smoking initiation,
IVW method was used to assess the causal effect and alcoholic consumption per week, BMI, and type 2
to determine heterogeneity. Multivariable MR-Egger diabetes). Sensitivity analyses revealed no horizontal
30
regression and MR Pleiotropy RESidual Sum and Outlier pleiotropy in the multivariable MR-Egger regression,
(MR-PRESSO) were employed to detect pleiotropy in except for MDD adjusted for smoking initiation
multivariable MR analysis. The MR-PRESSO method (intercept = −0.009, P = 0.04). However, no outliers were
31
can identify SNP outliers with multiple effects and provide detected using the MR-PRESSO method, and the global
the same estimated value as IVW after removing these test P-value was 0.18, with no evidence of horizontal
outliers. In addition, we tested conditional F-statistics to pleiotropy. Mild heterogeneity was observed when
32
evaluate the strength of the selected instruments using the adjusting for BMI and type 2 diabetes. For the F-statistics
MVMR R package. in multivariable MR, all conditional F-statistics exceeded
10, except for depression adjusted for triglycerides and
2.2.3. Two-step MR
BMI, which may suggest weak IVs (Figure 4, Tables S6,
To identify and evaluate the potential indirect influence S9, and S10).
of mediators on the total effect of depression on CP, we
conducted a two-step MR analysis. The selection of SNPs 3.2. Mediation through smoking initiation,
33
and sensitivity analyses were similar to those performed triglycerides, and type 2 diabetes
in the two-sample bidirectional MR. The specific steps We conducted a two-step MR analysis, and the results
in the two-step MR are as follows: the first step involves showed that the genetic susceptibility to depression was
determining the causal effect of depression (exposure) significantly associated with higher triglyceride levels,
34
on potential mediating factors through two-sample MR. smoking initiation, and an increased risk of type 2 diabetes
35
The second step estimates the causal effect of potential (Table S11). Among these, evidence of an indirect causal
36
mediating factors on CP (outcome) using both two-sample relationship between depression and CP was observed
and multivariable MR. The number of the IVs for potential through triglycerides (OR: 1.02; 95% CI: 1.01 – 1.04;
mediating factors selected is shown in Table S2. P = 0.004), type 2 diabetes (OR: 1.03; 95% CI: 1.01 – 1.06;
3. Results P = 0.030), and smoking initiation (OR: 1.06; 95% CI: 1.01
– 1.13; P = 0.046). As shown in Table 1, mediation analyses
Following the criteria for IV selection, 41 and 25 indicated that triglycerides, type 2 diabetes, and smoking
independent SNPs were ultimately chosen as the IVs for initiation accounted for 6.14%, 7.84%, and 17.06% of the
depression and CP, respectively (Tables S4 and S5). All effect from depression on CP, respectively. Moreover, no
F-statistics for the IVs exceeded 10, indicating strong horizontal pleiotropy was detected in any of the two-step
instrument variables (Tables S4, S5, and S6). MR analyses.

Volume 3 Issue 3 (2025) 58 doi: 10.36922/jcbp.5892