Figure 1 displayed the overview of the study design. In this study we performed a Mendelian randomization (MR) analysis to assess the causal relationship between 731 immunophenotypes and gynecologic malignancies. Three important assumptions must be met when choosing instrumental variables (IVs) ( Figure 1A ): Three requirements must be met for the genetic variations to be considered as IVs: (1) they must be strongly correlated with the exposure; (2) they cannot be linked to any confounders; and (3) the variants chosen should only influence the risk of the outcome by the risk factor independently not through other pathways (22).

The 731 immunophenotypes of peripheral blood were published by the genome wide association studies (GWAS) and are accessible to the general public through the GWAS database(GCST90001391-GCST90002121) (23). The original GWAS on immunophenotypes used information from 3,757 European individuals. It contains 4 trait types,7 panels, and 731 traits. Supplementary Table 1 provided an immunophenotype characterization. We selected four most common gynecologic malignancies including cervical cancer, ovarian cancer, endometrial cancer, and vulvar cancer as outcome. To obtain a more comprehensive conclusion of the causal links, we also took the carcinoma in situ into consideration. The GWAS summary data of the gynecologic cancer were accessed from the FinnGen study (24) (https://www.finngen.fi/en). Table 1 presented the detailed information of datasets used in this study. The analysis was based on summary-level data from large genome wide association studies that were made available in public. Therefore, ethical approval was not needed.

The IVs that met the strict significance threshold (P < 1 × 10⁻⁸) were chosen. SNPs with linkage disequilibrium were excluded concurrently (r² > 0.001, window size < 10,000 kb). In order to minimize the bias caused by weak IVs, SNPs with F-statistics < 10 were also eliminated. Next, we looked through and eliminated SNPs corresponding to confounders via the PhenoScanner website (25). The confounders included (age at menarche (26), trunk fat mass, body mass index (27), obesity, treatment with ovestin 0.1% vaginal cream, and treatment with estrogen product). Finally, 251independent SNPs were obtained as IVs for immunophenotypes.

In measuring the causal relationships between immunophenotypes and gynecologic malignancies, four MR methods (inverse variance weighted (IVW) (28), MR Egger (29), Weighted median (30), and Weighted mode) were utilized, including. The primary analysis was the IVW. The Benjamini-Hochberg, which regulates the false discovery rate (FDR), was used to modify multiple testing. The heterogeneity was evaluated using the Cochran’s Q test. MR-Egger intercept and leave one out analysis were used to determine horizontal pleiotropy. Immunophenotypes with adj.P value <0.05 were deemed to have a highly probable relationship with gynecologic malignancies(statistically significant), while those that displayed P value <0.05 after MR analyzes, but 0.05< adj.P value <0.2 were considered possible factors. After excluding IVs that exhibited pleiotropic effects, we performed the primary MR analysis once more. To investigate if exposure was directionally causal for the outcome, we applied the MR Steiger directionality test (31). All analyses were carried out in R software 4.3.1 utilizing the “Two Sample MR” and “Mendelian Randomization” packages.

For an insight into the relationship between immunophenotypes and four gynecologic cancers, we applied a tow-sample MR analysis. This study identified 252 independent SNPs linked to immunophenotypes ( Supplementary Table 2 ). By applying IVW methods, there were 50 highly probable immunophenotype traits (adj.P value <0.05, Tables 2 , 3 , Supplementary Table 5 ) and 65 possible immunophenotypes (P value <0.05, 0.05< adj.P value <0.2, Supplementary Tables 4 , 5 ) linked with gynecologic malignancies. When the highly probable immunophenotypes were classified in 7 panels (B cell, cDC, maturation stages of T cell, monocyte, myeloid cell, TBNK, Treg), 19 traits belonged to B cell, 8 from TBNK, 6 from cDC, monocyte, and maturation stages of T cell respectively, 4 from myeloid cell, and 2 from Treg. Sensitivity studies were performed to guarantee the robustness of the causal associations because IVW approaches are prone to weak IVs bias.

Table 2 and Figure 8 demonstrated that 32 pairs had highly probable causal effects (adj.P <0.05) between immunophenotypes and cervical cancer. The causal effect of immunophenotypes on adenocarcinomas of cervix was shown in Figure 2A . The B-cell panel showed the highest number of significant associations when compared to other panels, and the majority of B cells were protective factors against cervical cancer. In the study we conducted, B-cell activating factor receptor (BAFF-R) was the most frequently expressed molecule in our analysis. It expresses on several B cell subtypes, such as CD24⁺ CD27⁺ B cell, IgD⁺ CD24⁺ B cell, IgD⁺ CD24- B cell, IgD⁺ CD38^- B cell, memory B cell, and naive-mature B cell. However, CD20 on IgD+ CD38+ B cell (OR=1.887, 95%CI:1.078-3.306, P=0.026) had a significant association with the probability of cervical cancer ( Table 2 , Figure 2B ). In cDC panels, CD123 on plasmacytoid Dendritic Cell (OR=2.48, 95%CI:1.229-5.003, P=0.011), CD123 on CD62L⁺ plasmacytoid Dendritic Cell (OR=2.5, 95%CI:1.231-5.077, P=0.011), CD80 on plasmacytoid Dendritic Cell (OR=2.62, 95%CI:1.244-5.515, P=0.011), CD80 on CD62L⁺ plasmacytoid Dendritic Cell (OR=2.641, 95%CI:1.246-5.596, P=0.011) had a disadvantageous association with cervical cancer( Table 2 , Figure 2B ). The causal effect of immunophenotypes on Malignant neoplasm of cervix uteri was shown in Figure 3 .

No immunophenotype was found at the criterion “adj.P value <0.05” after FDR correction. Nevertheless, 18 immunophenotypes that fit the requirements of P value <0.05 and 0.05< adj.P value <0.2 are possible factors in our study ( Figures 4 , 9 ). B cells were also vital in endometrial cancer, similarly to how they were in cervical cancer. But all possible factors in B cell panel, namely, CD38 on IgD⁺ CD24^- B cell (OR=1.286, 95%CI; 1.038-1.594, P=0.021), IgD on IgD⁺ CD24- B cell (OR=1.09, 95%CI; 1-1.188, P=0.049), IgD on IgD⁺ CD38dim B cell (OR=1.096, 95%CI; 1.001-1.199, P=0.046), BAFF-R on IgD⁺ CD24⁺ B cell (OR=1.219, 95%CI; 1.003-1.483, P=0.047), BAFF-R on IgD- CD24- B cell (OR=1.249, 95%CI; 1.019-1.532, P=0.032), BAFF-R on IgD^- CD27^- B cell (OR=1.249, 95%CI; 1.02-1.529, P=0.032), BAFF-R on IgD- CD38⁺ B cell (OR=1.915, 95%CI; 1.12-3.273, P=0.018) were correlated with increased risk of endometrial cancer. In other panels, CD25⁺⁺ CD45RA⁺CD4⁺ T cell, CD28^- CD8⁺ T cell, CCR7 on naive CD8⁺ T cell, myeloid Dendritic Cell, Natural Killer cell, Basophil cell had positive links with endometrial cancer. However, CD62L on CD62L⁺ plasmacytoid Dendritic Cell (OR=0.831, 95%CI; 0.699-0.989, P=0.037) and CD14^- CD16⁺ monocyte %monocyte (OR=0.465, 95%CI; 0.227-0.953, P=0.036) were negatively associated with the risk of endometrial cancer.

We noticed 9 highly probable immunophenotype traits were connected to ovarian cancer in IVW test ( Figure 8 , Table 3 ). In the Monocyte panel, reduction in the risk of ovarian cancer was associated with HLA DR on CD14⁺ CD16- monocyte (OR=0.823, 95%CI:0.74-0.916, P=3.381E-04), HLA DR on CD14⁺ monocyte (OR=0.819, 95%CI:0.734-0.913, P=3.365E-04), and HLA DR on monocyte (OR=0.824, 95%CI:0.73-0.93, P=0.002). However, immune cells of different traits had opposing effects on ovarian cancer in TBNK and Myeloid cell panels. HLA DR⁺⁺ monocyte, HLA DR on CD33⁺ HLA DR⁺ CD14dim, and HLA DR on CD33^- HLA DR⁺ exhibited favorable effects. On the other hand, Monocytic Myeloid-Derived Suppressor Cells and CD45 on B cell were positively associated with the increasing risk of ovarian cancer. The causal effect of immunophenotypes on ovarian cancer was shown in Figure 5 .

The IVW analysis indicated that 9 highly probable immunophenotype traits had causal relationship with vulvar cancer ( Figure 8 , Table 3 ). Terminally Differentiated CD4⁺ T cell Absolute Count (OR=3.135, 95%CI:1.247-7.883, P=0.015) and HLA DR on B cell (OR=1.391, 95%CI:1.115-1.736, P=0.003) were correlated with increased vulvar cancer risk. Whereas HLA DR⁺⁺ monocyte %leukocyte (OR=0.443, 95%CI:0.248-0.793, P=0.006), CD20 on IgD⁺ CD24⁺ B cell (OR=0.326, 95%CI:0.139-0.765, P=0.010), CD4 on HLA DR⁺ CD4⁺ T cell (OR=0.239, 95%CI:0.099-0.579, P=0.002), HLA DR on CD14⁺ CD16^- monocyte (OR=0.69, 95%CI:0.543-0.876, P=0.002), HLA DR on CD14⁺ monocyte (OR=0.682, 95%CI:0.533-0.873, P=0.002), HLA DR on monocyte (OR=0.706, 95%CI:0.537-0.928, P=0.013), HLA DR on CD33⁺ HLA DR⁺ CD14^- (OR=0.642, 95%CI:0.466-0.884, P=0.007) were related to a reduced risk of vulvar cancer. In addition, we also screened 5 possible immunophenotypes for vulvar cancer. Figure 6 and Supplementary Table 3 presented the detailed data.

There are 65 possible immunophenotypes for gynecologic cancer in situ, but we found only one highly probable immunophenotype trait after FDR correction. There are 65 possible immunophenotypes for gynecologic cancer in situ, but we found only one highly probable immunophenotype trait after FDR correction ( Figure 10 ). Terminally Differentiated CD4^-CD8^- T cell (OR=8.095, 95CI%: 2.206-29.699, adj.P= 0.039) was significantly positive link with carcinoma in situ of vulva risk.

In carcinoma in situ of cervix uteri, 26 causative links between the immune cells and cervical carcinoma in situ have been identified ( Figure 7A , Supplementary Table 5 ). Compared to other panels, TBNK had the greatest number of possible associations. CD8dim T cell was linked in a higher incidence of cervical carcinoma in situ, but NKT cell, HLA DR⁺ CD4⁺ T cell, and HLA DR⁺ NK cell were protective immune factors. Treg cells also play an important role. CD39⁺ CD8⁺ T cell %CD8⁺ T cell (OR=1.14, 95%CI:1.012-1.284, P=0.032), CD3 on activated CD4 regulatory T cell (OR=1.103, 95%CI:1.007-1.208, P=0.035), CD3 on activated & secreting CD4 regulatory T cell (OR=1.1, 95%CI:1.006-1.204, P=0.036), CD3 on CD39⁺ CD4⁺ T cell (OR=1.112, 95%CI:1.016-1.216, P=0.021) were linked in a higher incidence of cervical carcinoma in situ.

Endometrial carcinoma in situ was revealed to be causally related to 28 possible immunophenotypes ( Figure 7B , Supplementary Table 5 ). The Treg panel’s characteristics were still prominent in endometrial cancer in situ, similar to carcinoma in situ of the cervix uteri.CD3 on T cell (OR=0.463, 95%CI:0.26-0.827, P=0.009), CD3 on activated CD4 regulatory T cell (OR=0.576, 95%CI:0.377-0.882, P=0.011), CD3 on CD39⁺ activated CD4 regulatory T cell (OR=0.538, 95%CI:0.347-0.835, P=0.006), CD3 on activated & secreting CD4 regulatory T cell (OR=0.582, 95%CI:0.382-0.884, P=0.011), CD3 on CD39⁺ CD4⁺ T cell (OR=0.598, 95%CI:0.394-0.909, P=0.016), CD3 on CD28⁺ CD4⁺ T cell (OR=0.482, 95%CI:0.276-0.842, P=0.01), CD3 on CD28⁺ CD45RA^- CD8⁺ T cell (OR=0.416, 95%CI:0.215-0.804, P=0.009), and CD8 on CD28⁺ CD45RA⁺ CD8⁺ T cell (OR=0.621, 95%CI:0.415-0.931, P=0.021) were inversely correlated with the incidence of endometrium carcinoma in situ. In Myeloid cell panel, CD45 on Immature Myeloid-Derived Suppressor Cells (OR=1.65, 95%CI:1.016-2.679, P=0.043) was linked in a higher incidence risk, while CD33⁺ HLA DR⁺ CD14dim, Granulocytic Myeloid-Derived Suppressor Cells, CD33 on CD14⁺ monocyte, CD33 on CD33^dim HLA DR⁺ CD11b⁺, CD33 on CD33dim HLA DR⁺ CD11b^-, and CD33 on basophil were protective factors for carcinoma in situ of endometrium. We also found that there were causal links between B cells, T cell and cDC.

In carcinoma in situ of vulva, we identified 11 possible immunophenotypes ( Figure 7C , Supplementary Table 5 ). HLA DR⁺⁺ monocyte, CD20 on IgD⁺ CD24⁺ B cell, CD4 on HLA DR⁺ CD4⁺ T cell, HLA DR on CD14⁺ CD16^- monocyte, HLA DR on CD14⁺ monocyte, HLA DR on monocyte, and HLA DR on CD33⁺ HLA DR⁺ CD14^- had a negative association with vulvar carcinoma in situ. The following factors might raise the incidence of vulvar cancer in situ: CD25⁺⁺ CD45RA⁺ CD4 not regulatory T cell, Terminally Differentiated CD4⁺ T cell, CD16⁺ monocyte, HLA DR⁺ Natural Killer cell, CD40 on CD14⁺ CD16⁺ monocyte, and CD45RA on naive CD8⁺ T cell.

Sensitivity analysis was performed to ensure the robustness of the causal evaluation because IVW methods are prone to weak instrumental bias. The highly probable or possible immunophenotypes showed no evidence of horizontal pleiotropy when examined using the MR-Egger intercept method (all P > 0.05). Cochran’s Q statistic test did not reveal any evidence of heterogeneity (all P > 0.05). The MR Steiger directionality test additionally demonstrated the causative links between immune cells and gynecologic malignancies. In our results, reverse causality was not observed. ( Supplementary Tables 7 ).