(1) Characteristics of study participants
A total of 281 patients completed a 6-week antidepressant treatment course. Among these patients, 205 achieved a response and the response rates about 72.9%. The demographic and clinical characteristics of patients in the responder and non-responder groups are shown in Table 1. There were no significant differences between the 6-week responder and non-responder groups in sex, age, drugs used, years of education, or family history of mood disorders. However, the baseline HDRS-17 score was significantly different between these two groups (t = 2.891, P=0.004).
A total of 275 patients completed 8-week antidepressant treatment. Among these patients, 144 achieved remission and the remission rates about 52.4%. There were no significant differences in age, number of years of education, family history, baseline HDRS-17 score, or antidepressant agents used between remission and non-remission groups (all P > 0.05), while the proportion of male patients and number of episodes were significantly higher in the non-remission group than the remission group (t = 2.381, P=0.018 and t = −1.983, P=0.049, respectively), as shown in Table 1 and 2.
(2) Allele frequency and linkage disequilibrium
Among the eight SNPs of the three genes investigated, two (ApoA4 rs5101 and rs675) were eliminated as they had MAF < 5%, while the remaining six SNPs were subjected to further statistical analyses (table S1). LD analysis showed that two SNPs (rs1801133, rs1801131) of the MTHFR gene were in near 100% LD (D' = 1.0, r2 = 0.177), while two other SNPs (rs405509, rs439401) of the ApoE gene were in strong LD (D' = 0.961, r2 = 0.505). The other SNPs showed no LD.
(3) Single polymorphism associations for the MTHFR, ApoE and ApoA4
Analysis of single-locus effects revealed that genotypes of the SNPs: rs1801131and rs1801133 (MTHFR) had no significant association with antidepressant response in 6 weeks (The detail negative results see the supplement material TableS2, S3). However, MTHFR rs1801133 TT genotype was related to the efficacy of antidepressants, and its distribution frequency is significantly higher in the non-remission group than in the remission group (group overall: χ2 = 6.328, P = 0.012; female subgroup: χ2 = 4.145, P = 0.042; SSRI subgroup: χ2 = 5.808, P = 0.016), but the results did not withstand permutation correction (The detail negative results see the supplement material Tables S4 and S5).
The ApoE rs405509 C allele was significantly associated with the efficacy of antidepressants at 6 weeks (Overall group: χ2 = 6.27, P = 0.012, Male subgroup: χ2 = 8.445, P = 0.004) ;; SNRI subgroup: χ2 = 6.707, P = 0.0096), compared with the A allele, the antidepressant efficacy of the C allele carrier is worse (Overall group: OR = 0.6, 95% CI = 0.4-0.9; Male subgroup: OR = 0.41 95% CI = 0.22-0.75; SNRI subgroup: OR = 0.42, 95% CI = 0.22-0.82), in which the results only in the male subgroup withstood 1000 permutations testing (P* < 0.05); The C allele was also compared with the 8-week SNRI subgroup The worse antidepressant efficacy was related (P = 0.031, OR = 0.70), but the results did not withstand permutation testing. The rs405509 AA genotype was significantly associated with the efficacy of antidepressants at 6 weeks (Overall group: χ2 = 7.41, P = 0.006; Male subgroup: χ2 = 9.77, P = 0.002; SNRI subgroup: χ2 = 6.41, P = 0.011), compared with AC genotype, AA genotype carriers have better antidepressant effect (Overall group: 95% CI =0.48(0.28-0.84); Male subgroup: OR = 0.28 95% CI = 0.12-0.66; SNRI subgroup: OR = 0.34, 95% CI = 0.13-0.87). The results in the Overall group (P* = 0.04) and SNRI subgroup (P* = 0.009) withstood permutation testing. (Table 3)
The APOE rs7412 G allele was associated with better antidepressant efficacy in the male subgroup (6-week male subgroup: P = 0.009, OR = 3.15; 8-week male subgroup: P = 0.0109, OR = 3.92), but none of the results did not withstand permutation testing. There was no significant correlation between the ApoE rs7412 genotype and the efficacy of antidepressants (all P> 0.05). There was no significant correlation between APOE rs439401 allele and antidepressant efficacy (all P > 0.05). The rs439401 GG genotype was associated with better antidepressant efficacy in the 8-week SSRI subgroup (χ2=4.313, P=0.038, OR=2.07, 95% CI = 0.81-5.2), but the results did not withstand permutation testing (The detail negative results see the supplement material Table S6, S7, S8 and S9).
The ApoA4 rs5092 G allele was associated with antidepressant response, of which the antidepressant effect of G allele carriers was poor in 6-week male subgroup(χ2 = 4.334, P = 0.037, OR = 0.55, 95% CI = 0.31-0.97), 6-week SNRI subgroup(χ2 = 7.241, P = 0.007, OR = 0.42, 95% CI = 0.22-0.80) and 8-week female subgroup(χ2 = 4.014, P = 0.045, OR=0.64 95% CI = 0.41–0.99) ), the result withstood permutation testing (6-week male subgroup: P* = 0.03, 6-week SNRI subgroup: P* = 0.005, 8-week female subgroup: P* = 0.03) (Table 3). Furthermore, the ApoA4 rs5092 GG genotype was significantly associated with antidepressant efficacy in the 6-week male subgroup (χ2 = 4.059, P = 0.034) and SNRI subgroup (χ2 = 6.964, P = 0.008). Compared to the AA and AG haplotypes, the GG haplotype was associated with reduced likelihood of a good response (male subgroup, OR = 0.26, 95% CI = 0.08–0.87; SNRI subgroup, OR = 0.13, 95% CI = 0.02–0.65), but only the SNRI subgroup withstood permutation correction (P*= 0.019) (Table 3). The other negative results see the supplement material Table S10, S11, S12 and S13.
(4) Haplotype associations for the MTHFR and ApoE
We examined the associations of haplotypes derived from the SNPs in MTHFR and ApoE with antidepressant response, limiting our analysis to haplotypes with a frequency of 5%. Table 4 shows that the haplotype (C-A) in MTHFR (rs1801133 and rs1801131) was significantly associated with antidepressant response in the 8-week antidepressant group overall (χ2 = 11.39, P = 0.0007), male subgroup (χ2 = 8.767, P = 0.003), and SNRI subgroup (χ2 = 10.51, P = 0.001). In comparison to the T-A haplotype and C-C haplotype, the C-A haplotype was associated with increased likelihood of good remission in the group overall (OR = 1.718, 95% CI = 1.178–2.505), and in the male subgroup (OR = 1.971, 95% CI = 1.088–3.572) and SNRI subgroup (OR = 2.251, 95% CI = 1.24–4.085), all of the results outlined above withstood permutation testing (overall: P* = 0.02, male subgroup: P* = 0.012, SNRI subgroup: P* = 0.002) (Table 4). The haplotype in MTHFR (rs1801133 and rs1801131) was no significant association with antidepressant response in the 6-week (all P >0.05), and the detail results were provided in the supplement material S14, S15).
Another analysis of haplotype effects demonstrated that haplotype (G-A) of ApoE (rs7412 and rs405509) was significantly associated with antidepressant response in the 6-week male subgroup (χ2 = 8.687, P = 0.003) In comparison to the A-A haplotype, the G-A haplotype was associated with increased likelihood of a better response (OR = 1.24, 95% CI = 0.12–12.9). Haplotype (G-C) in ApoE (rs7412 and rs405509) was significantly associated with antidepressant response in the 6-week SNRI subgroup (χ2 = 8.24, P=0.0041). In comparison with the A-A haplotype, the G-C haplotype was associated with increased likelihood of a better response (OR = 1.04, 95% CI = 0.19–5.64). All of the results outlined above withstood permutations testing (male subgroup: P* < 0.05, SNRI subgroup: P* = 0.049) (Table 5).
We also found that in the overall group, haplotypes G-A and G-C (rs7412 and rs405509), haplotypes CG (rs405509 and rs439401), and haplotypes GCG (rs7412, rs405509, and rs439401) was significantly associated with antidepressant response in the 6-week (rs7412-rs405509: G-A, χ2 = 5.046, P= 0.025; G-C, χ2 = 4.313, P = 0.038; rs405509-rs439401: χ2 = 5.13, P = 0.024; rs7412-rs405509-rs439401: χ2 = 3.907, P = 0.048); In the male subgroup, haplotypes A-G, C-G (rs405509 and rs439401), and haplotypes GAG (rs7412, rs405509, and rs439401) were associated with antidepressant response (rs405509-rs405509: A-G, χ 2 = 5.876 , P = 0.015; C-G, χ2 = 6.136, P = 0.013; rs7412-rs405509-rs439401: GAG, χ2 = 5.842, P = 0.016); In the SNRI subgroup, haplotype G-A(rs405509 and rs439401) ,A-A, C-G, and GAG, GCG, (rs7412, rs405509, and -rs439401) was significantly associated with antidepressant response in the 6-week , (rs7412-rs405509: χ2 = 7.658, P = 0.006; rs405509-rs439401: AA χ2 = 4.02, P = 0.045, C-G, χ2 = 5.392, P = 0.02; rs7412-rs405509-rs439401: G-A-A, χ2 = 4.367, P = 0.037, G-C-G, χ2 = 6.517, P = 0.011). However, all of the results outlined above did not withstand permutation testing (see Table S16).
In the SNRI subgroup, the haplotype G-A (rs7412 and rs405509), and C-G (rs405509 and rs439401) of ApoE was significantly associated with worse antidepressant response in the 8-week (rs7412-rs405509: P = 0.044, OR = 0.91, rs405509-rs439401: P = 0.047, OR = 0.55), but the results outlined above did not withstand permutations (see Table S17).