Results
Patient characteristics. Sixty-eight healthy Chinese Han women for pre-pregnancy examination were included in this study, as described in our previous report(21). Table S1 showed in summary the demographic characteristics of the 68 subjects. The mean age of the subjects was (29.9 ± 3.9) years, where 36 (52.9%) were younger than 30 years. The Body Mass Index (BMI) of 59 subjects (86.8%) was below 24 kg/m2. The average AAM was (14 ± 1.6) years old, where 26 (38.2%) experienced AAM below the age of 14 years. The average menstrual cycle was 30 days, where 10 females (14.7%) had abnormal cycles, among of which two subjects (2.9%) had short menstrual cycles (less than 24 days) and 8 ones (11.8%) had prolonged menstruation (more than 35 days). In addition, 40 females (58.8%) had at least one gravidities, and 33 ones (48.5%) were parous. Twenty (29.4%) had one miscarriage, and nine (13.2%) have had more than two miscarriage.
There were 35 pregnancies after 15 months follow-up. Among of these subjects, the body weight of 10 ones (41.2%) increased more than 19.98 kg at early pregnancy. And the median of TSH, FT4 and TPO-ab of pregnancy subjects at early pregnancy were 1.9 mIU/ml, 10.4 pmol/L, and 32.1 IU/ml, respectively. During the mid-trimester pregnancy, the median of fast glucose, triglyceride, cholesterol , HbA1c, and systolic/diastolic blood pressure was 4.5 mmol/L, 2.1 mmol/L, 6.1 mmol/L, and 117/66 mmHg, respectively.
In these 35 gravidas, there were 20 healthy pregnancies, 7 threatened miscarriages and 8 spontaneous miscarriages. No chromosome abnormality and congenital deformity of the fetuses was found in these miscarriages.
Early AAM is associated with pre-pregnancy Tfh1/Tfh2 ratio. We analyzed the correlations between AAM and T cells, γδ T cells and their subsets according to the method as described(21). No significant correlations was found between AAM and CD3+, double positive T (CD4+CD8+) and double negative (CD4-CD8-) T cells, T helper cells (CD4+, also known as Th cells) and subsets Th1/2/17, and cytotoxic T cells (CD8+, also known as Tc cells) and subsets Tc1/2/17 (all P values >0.05). At the same time, no significant correlations were found between AAM and CD4/CD8 memory cells such as Naïve, central memory (CM), effect memory (EM), and EM CD45RA+ (EMRA). No significant correlations were found between AAM and γδ T cells either (all P values > 0.05).
Among T follicular helper cells (CD3+CD4+CXCR5+, also known as Tfh cells), Tfh1 (CD4+CXCR5+CXCR3+CCR4-) were significantly negatively correlated with AAM (r =-0.273,P =0.027), wherever Tfh2 (CD4+CXCR5+CXCR3-CCR4+) were positively associated with AAM (Spearman r =0.342,P =0.005), as shown in Figure 1B. Further, the Tfh1/Tfh2 ratio also showed significantly negative correlated with AAM (Spearmanr =-0.283, P =0.019). There was no discernible correlation between Tfh17 (CD4+CXCR5+CXCR3-CCR4-CCR6+) and AAM (Spearman r =-0.137, P =0.265).
Pre-pregnancy Tfh1/Tfh2 ratio is associated with TSH at early pregnancy. The pregnancy status of the subjects were followed-up by interview until December 2019. After a 15-month follow-up, there were 35 pregnancies, including 20 healthy pregnancies, 7 threatened miscarriages and 8 spontaneous miscarriages.
We analyzed the correlation between pre-pregnancy Tfh1/Tfh2 ratio and some physical examination indexes during pregnancy. We found that the ratios of Tfh1/Tfh2 at pre-pregnancy examinations were significantly positively associated with TSH at early pregnancy (Spearmanr =0.367, P =0.046, Figure 2 ).
However, there no significant associations between pre-pregnancy Tfh1/Tfh2 ratio and either of the following indexes: weight gain, FT4 and TPO-ab at early pregnancy; and BMI, blood pressure, fast glucose, triglyceride, cholesterol, and HbA1c during mid-trimester pregnancy (allP values >0.05, as shown in Table S2 ).
Pre-pregnancy Tfh1/Tfh2 ratio was associated with an increased risk and a shorter time to spontaneous miscarriage. Consistent with previous reports(10), the average AAM of threatened miscarriages and spontaneous miscarriages was both early than healthy pregnancies (bothP values <0.05, Table S3 ).
Therefore, we analyzed the relationship between pre-pregnancy Tfh1/Tfh2 ratio and miscarriages. Firstly, Manny-Whitney U tests were used to compare immune cells between healthy pregnancies and threatened or spontaneous miscarriages. We found that Tfh1/Tfh2 ratio was higher in spontaneous miscarriages than the ratio in healthy pregnancies (median 0.288 vs. 0.179, Manny-Whitney U=-2.288, P =0.021, Figure s1 ), but not in threatened miscarriages.
We further studied the association of spontaneous miscarriage rates with the Tfh1/Tfh2 ratio and AAM during the study period in 35 gravidas. Rate of spontaneous miscarriages were analyzed by the Kaplan-Meier estimate and Cox regression analyses. The log-rank test was used to compare spontaneous miscarriages curves. We found that the women with AAM<14 years had a higher risk of spontaneous miscarriages (adjusted RR=17.28, 95% CI=2.04-146.75, P =0.009, Table 1 ), and a shorter spontaneous miscarriage time (45 days vs. 185 days, log-rank P =0.036, Figure 3A ). The adjusted confounding factors included age, previous times of spontaneous abortion and BMI. Moreover, the women with the Tfh1/Tfh2 ratio≥0.209 (median) also had a higher risk of spontaneous miscarriages (adjusted RR=12.25, 95% CI=1.72-87.46, P =0.013, Table 1 ) and a short time to the spontaneous miscarriage (42 days vs. 115 days, log-rank P =0.038,Figure 3B ). Our results implied a possible logical connection with earlier AAM, higher pre-pregnancy Tfh1/Tfh2 ratio and spontaneous miscarriage.
However, no distinguishable association between mothers’ menarche age and the gestational weeks and birth weight of newborns were found in this study.
Genetic factors of AAM contributes to Tfh cells. To test the causal relationship and shared genetic factors between Tfh cells and AAM, we performed two sample MR tests by using publicly available genetic summary data of Genome-wide association studies (GWAS) from 176008 subjects in UK Biobank (http://www.nealelab.is/blog/2017/9/11/details-and-considerations-of-the-uk-biobank-gwas ) and 497 female participants from the UK Adult Twin Register (TwinsUK) (22) in MR-base(24).
At a GWAS threshold of statistical significance (P <5×10-8; linkage disequilibriumR 2<0.1), we found that 84 AAM-related SNPs were significantly associated with PD1- naïve Tfh cells by Weighted median method (β=0.38±0.19, P =0.040,Figure S2 ) and by Weighted mode (β=0.62±0.31, P =0.045). Moreover, these SNPs can explain 19% of variance in PD1- Tfh naïve cells (DirectionalityP =4.28×10-10). However, there were no significant causal relationship between AAM and two other types of naïve Tfh cells (CCR4-CXCR3- and CCR4+CXCR3+), CD4+Tfh cells and CD8+ Tfh cells in MR-base (All Pvalues > 0.05).
In MR-Base, there were another two AAM database: ReproGen consortium studies(25) and MRC-IEU consortium studies (http://www.ewascatalog.org ). However, there were no significant association between AAM and Tfh cells by using GWAS Data from ReproGen and MRC-IEU consortium studies. These results might due to the small samples with outcomes in TwinsUK dataset, and the genetic background of TwinsUK might be more different to ReproGen and MRC-IEU consortium studies than to UK Biobank.
We further performed a leave-one-out analysis to assess whether the MR estimate is driven or biased by a single SNP. Using this method, we found that rs1516883 and rs1933801 were the major share genetic factors between AAM and PD-1- naïve Tfh cells (also shown asFigure S2 ). These two SNPs were located in chromosome region 9p12 (LINC01505, TAL2 and TMEM38B ) and near geneLIN28B respectively, both are reported AAM loci(25-28).