3.1 Dyslipidemia in EBC dataset
3.1.1 Participants description
A total of 5917 EBC patients were included in the EBC women data set. The median age was 47 years old [IQR, 42.00, 55.00], and the median BMI was 22.83 [20.83, 24.97]. The proportion of postmenopausal and non-menopausal female patients were 38.1% and 61.9% respectively. The baseline blood lipid levels were as follows: TC(Median (IQR) , 4.66 [4.12, 5.29] mmol/L), TG(Median (IQR), 1.10 [0.82, 1.54] mmol/L), HDL-C(Median (IQR), 1.56 [1.32, 1.84] mmol/L) , LDL-C(Median (IQR), 2.59 [2.16, 3.10] mmol/L).
3.1.2 Prevalence of Dyslipidemia and Risk Factors
At the initial diagnosis of EBC women data set, 978 cases (16.5%) had dyslipidemia, and 75 / 978 cases (7.7%) reached the high-risk blood lipid level of ASCVD. As shown in Table 2, the prevalence of dyslipidemia in the higher age group was higher than that in the lower age group (P < 0.001). The prevalence of dyslipidemia in each age group was 18-34 years old (6.4%), 35-44 years old (9.7%), 45-54 years old (16.7%), 55-64 years old (26.3%) and 65 years old(28.7%). The proportion of blood lipid reaching the high level of ASCVD was also statistically significant between age groups (P = 0.012). The proportion of dyslipidemia was also different between different BMI groups (P<0.001), and the proportion of dyslipidemia in Obese group (22.9%) or Overweight group (28.3%) was higher than that in low BMI (8.3%) or normal BMI group (12.9%). The abnormal proportion of blood lipids in menopausal patients (25.7%) was higher than that in non-menopausal patients (10.9%), and there was a similar trend for blood lipids to reach the high-risk level of ASCVD (menopause (10.0%) vs non-menopause (4.2%), P = 0.001). There was no significant difference in the prevalence of dyslipidemia among Subtypes, Stage and each hormone receptor group (ER, PR, Her2 and Ki67).
Age, BMI, menopause status, stage, ER, PR, HER2, Ki67 were included to construct binary multivariate logistic regression model. As shown in Figure 1, the risk of dyslipidemia was increased in both Overweight and Obese group compared with low BMI levels(Odds ratio [OR], Overweight: 2.69 [95% CI, 1.57-4.90], Obese: 2.27 [95% CI, 1.40-3.90]). Compared with the younger group (less than 35 years old), the risk of dyslipidemia was increased in the 55-64 and 65- age groups(Odds ratio [OR], 55-64: 1.86 [95% CI, 1.08-3.40] , 65-: 2.25 [95% CI, 1.14-3.90]). The prevalence of dyslipidemia in postmenopausal group was higher than that in non-menopausal group, and its prevalence of dyslipidemia increased(Odds ratio [OR], 1.95 [95% CI, 1.51-2.50]).
3.2 Dyslipidemia in EBC Endocrine Dataset
3.2.1 Dyslipidemia
In this study, 1883 female patients with normal blood lipid before endocrine therapy were followed up for 60 months to monitor their blood lipid levels. Among them, there were 1258 non-menopausal cases, the median age was 43.00 years old [IQR, 39.00-47.00], the median BMI was 22.22 [IQR, 20.45-24.14], 298 cases had dyslipidemia within 60 months, and the median follow-up time was 24.40 months [IQR, 0.03, 60.00]. There were 625 postmenopausal patients, the median age was 57.00 years old [IQR, 52.00-62.00], the median BMI was 23.44 [IQR, 21.23-25.63], 207 patients had dyslipidemia within 60 months, and the median follow-up time was 20.06 months [IQR, 0.03, 60.00]. The incidence of dyslipidemia in one year in postmenopausal patients was significantly higher than that in non-menopausal patients (25.2 % [95%CI, 21.6 %-28.7 %] vs 16.8% [95%CI, 14.6%-18.9%], P<0.001; Supplement Table 1 and Supplement Figure 1). The 5-year risk of dyslipidemia in postmenopausal patients was also significantly higher than that in non-menopausal patients (Adjusted HR [95%CI], 1.29 [ 1.04 – 1.59], P=0.0186; 42.6 % vs 32.6%, P<0.001; Supplement Table 1 and Supplement Figure 1).
3.2.2 Abnormalities in Four Blood Lipid Indexes among Premenopause EBC
We further analyzed the effect of specific endocrine therapy on the incidence of single index of dyslipidemia. The non-menopausal group received three types of endocrine therapy modes: SERM (77.9%), OFS+AI(15.0%), OFS+SERM(7.1%).The specific plans were: TAM(74.7%)and TOR(3.2%);OFS+TAM(6.9%)and OFS+TOR(0.2%);OFS+ANA(7.7%),OFS+LET(5.3%),OFS+EXE(2.0%). Before endocrine therapy, the blood lipid levels were TC (4.74 [4.22, 5.25] mmol/L),TG(1.28 [0.97, 1.64] mmol/L),HDL-C (1.45 [1.27, 1.68] mmol/L)and LDL-C(2.75 [2.31, 3.16] mmol/L). There were significant differences in the risk of TC and LDL-C abnormalities among the three groups(TC: P<0.001, LDL-C: P<0.001;Figure 2). The results of Pairwise log-rank test showed that compared with SERM group, the 1-year abnormal incidence of TC and LDL-C in OFS + AI Group was significantly higher(TC: OFS+AI vs SERM, 22% vs 3.5%, P<0.001; LDL-C: OFS+AI vs SERM, 12.2% vs 1.4% , P<0.001; Supplement Table 2 and Supplement Table 4);the 5-year abnormal incidence of TC and LDL-C in OFS + AI Group was also significantly higher(TC: OFS+AI vs SERM, 22.0% vs 3.5%, P<0.001; LDL-C: OFS+AI vs SERM, 12.2% vs 1.4% , P<0.001; Supplement Table 2 and Supplement Table 4). In the subgroup analysis of specific drugs, we found that compared with TAM group, the 5-year incidence of TC and LDL-C in OFS+ANA or OFS+LET was significantly increased, and the difference between OFS + ANA and OFS + LET was not significant(Supplement Table 2 and Supplement Table 6). In order to determine the relative risk of different treatment modes or drugs in the occurrence of TC abnormality and LDL-C abnormality, we constructed a multivariate COX regression model, the factors with significant single factor COX regression P value between the corresponding groups were included for correction, and it was found that the OFS + AI group had a higher risk of LDL-C abnormality than the SERM group (Adjusted HR [95%CI], 6.71 [ 3.16 – 14.26], p<0.001)and a higher risk of TC abnormality (Adjusted HR [95%CI], 3.50 [ 1.74 - 7.02], p<0.001). Compared with TAM, OFS + ANA or OFS + LET had higher probability of LDL-C abnormal risk(OFS+ANA vs TAM: Adjusted HR [95%CI], 10.86 [ 4.59 - 25.71], p<0.001; OFS+LET vs TAM: Adjusted HR [95%CI], 3.66 [ 1.01 - 13.30], p=0.049)and TC abnormal risk(OFS+ANA vs TAM: Adjusted HR [95%CI], 4.05 [ 1.72 - 9.55], p=0.001; OFS+LET vs TAM: Adjusted HR [95%CI], 4.66 [ 2.02 - 10.72], p<0.001), and the others are shown in Table 3. However, there was no statistical difference in TG and HDL-C between different endocrine therapy groups in premenopausal patients(Supplement Figure 2).
3.2.3 Abnormalities in Four Blood Lipid Indexes among Menopause EBC
For the menopausal group, SERM (26.2%) and AI (73.8%) were used as endocrine therapy, the specific therapeutic drugs were: TAM(25.1%)and TOR(1.1%);ANA(38.2%),LET(29.8%)and EXE(5.8%). Before endocrine therapy, the blood lipid levels were: TC (4.98 [4.44, 5.45] mmol/L),TG(1.42 [1.09, 1.72] mmol/L),HDL-C (1.41 [1.23, 1.65] mmol/L)and LDL-C(2.93 [2.50, 3.37] mmol/L). Similar to the premenopausal group, there were significant differences in the risk of abnormal TC and LDL-C between SERM therapy and AI therapy(TC: P<0.001, LDL-C: P=0.002;Figure 3). The incidence of 1-year abnormal TC in AI group was higher than that in SERM group(AI vs SERM, 13.9% vs 4.5%, P=0.002; Supplement Table 3 and Supplement Table 5), and the incidence of abnormal TC in 5 years was also higher than that in SERM group(AI vs SERM, 21.9% vs 8.2%, P<0.001; Supplement Table 3 and Supplement Table 5). Compared with SERM group, the 1-year incidence of abnormal LDL-C in AI group did not increase significantly(AI vs SERM, 8.6% vs 2.5%, P=0.095), but the 5-year abnormal risk increased significantly(AI vs SERM, 13.3% vs 2.5%, P=0.002; Supplement Table 3 and Supplement Table 5). Similarly, in the subgroup analysis of specific drugs, we found the same trend as premenopausal : the incidence of abnormal TC and LDL-C in 1 year and 5 years of OFS + ANA or OFS + LET was significantly higher than that in TAM group, and the difference between OFS + ANA and OFS + LET was not significant(Supplement Table 3 and Supplement Table 7). Multivariate COX regression model showed that AI group had higher risk of abnormal LDL-C(Adjusted HR [95%CI], 3.92 [ 1.41 - 10.91], p=0.009)and abnormal TC(Adjusted HR [95%CI], 3.15 [ 1.58 - 6.29], p=0.001;Table 3)than SERM group. Compared with TAM, ANA or LET had higher probability of abnormal LDL-C (ANA vs TAM: Adjusted HR [95%CI], 6.17 [ 1.87 - 20.37], p=0.003; LET vs TAM: Adjusted HR [95%CI], 5.00 [ 1.46 - 17.07], p=0.010)and TC abnormal (ANA vs TAM: Adjusted HR [95%CI], 4.07 [ 1.92 - 8.62], p<0.001; LET vs TAM: Adjusted HR [95%CI], 2.77 [ 1.25 - 6.10], p=0.012). There was no statistical difference in TG and HDL-C between different endocrine therapy groups in postmenopausal patients(Supplement Figure 3).