Patient demographics and baseline characteristics
The overall study flow diagram is shown in Figure 1. In total, 17,698 patients were separated into five groups: POSEIDON group 1 (younger unexpected poor responders, n=4470), POSEIDON group 2 (older unexpected poor responders, n=2270), POSEIDON group 3 (younger expected poor responders, n=1110), POSEIDON group 4 (older expected poor responders, n=1095), and group 5 (control group, n=8753). Patients in the control group exhibited a sufficient ovarian reserve (AFC ≥ 5) and a sufficient number of oocytes retrieved (> 9) during the first ovarian retrieval cycle. Patient demographics and baseline characteristics are shown in Table 1.
When patients were grouped according to POSEIDON criteria (Table 1), the average ages of the older patients in group 2 (38.01±2.83) and group 4 (39.37±3.38) were notably higher than those of younger patients in group 1 (30.21±2.76), group 3 (30.32±2.71), and the control group (30.44±3.85). Consistent with grouping criteria, the average AFC of expected poor responders in group 3 (2.19±1.65) and group 4 (2.51±1.32) was significantly lower than that of group 1 (11.17±5.77), group 2 (8.99±4.09), and control group patients (15.24±6.21). Average numbers of oocytes retrieved during the first cycle were highest in the control group (16.84±6.57), followed by group 3 (7.07±6.67), group 1 (5.97±2.31), group 2 (5.27±2.34), and group 4 (3.29±3.05). Owing to heterogeneity among low-prognosis women, types of infertility, primary etiology, and BMI all differed significantly among these five groups (P<0.05). Furthermore, with increasing age, the duration of infertility rose group 1 (2.93±2.42) to group 3 (3.14±2.46), group 2 (4.08±3.96), and group 4 (4.32±4.56). Gravidity, parity, and year of treatment also differed significantly among these groups (P<0.05).
Ovarian stimulation characteristics and pregnancy outcomes
As expected, patients in group 4 (older expected poor responders) underwent the most oocyte retrieval cycles (1.97 ± 1.33), followed by patients in group 3 (younger expected poor responders, 1.55 ± 0.94), group 2 (older unexpected poor responders, 1.54 ± 0.89), and group 1 (younger unexpected poor responders, 1.33 ± 0.64), as shown in Table 2. Patients in the control group underwent the fewest oocyte retrieval cycles (1.11 ± 0.38) (P<0.05). However, the number of retrieved oocytes was smallest in group 4 (3.29±3.05), and increased in group 2 (5.27±2.34), group 1 (5.97±2.31), and group 3 (7.07±6.67), with the greatest number of oocytes having been retrieved for patients in the control group (16.84 ± 6.57) (P<0.05). As such, the oocyte output rate (number of oocytes retrieved / AFC x 100%)(19) was highest in group 3 (younger expected poor responders, 199.21%), followed by group 5 (normal or high responders, 109.28%), group 4 (older expected poor responders, 99.31%), group 2 (older unexpected poor responders, 62.96%), and group 1 (younger unexpected poor responders, 58.02%). Patients in the control group also exhibited the highest number of 2PN (10.39±5.09) and good quality embryos (5.32±3.78) per oocyte retrieval cycle, followed by patients in group 3 (4.47±4.26; 2.38±2.59), group 1 (3.86±2.08; 2.15±1.71), group 2 (3.44±1.95; 1.98±1.55), and group 4 (2.16±2.13; 1.27±1.39) (P<0.05).
Table 2 displays the CLBRs within 2 and 5 years in each group with corresponding 95% CIs. When using the optimistic method, the CLBRs over 2 years were 75.6%, 55.5%, 67.2%, 35.7%, and 76.2% in groups 1-5, respectively. Using the conservative method, the CLBRs over 2 years in groups 1-5 were 63.7%, 42.1%, 55.3%, 25.9%, and 70.5%, respectively. When the CLBR timeframe was extended to five years, similar decreasing trends were observed from group 5 (optimistic: 79.3%; conservative: 72.5%) to group 1 (80.8%; 66.0%), group 3 (75.0%; 58.4%), group 2 (60.3%; 43.6%), and group4 (41.4%; 27.2%). However, there may be differences in the extent of these CLBRs among these groups over the entirety of this 5-year period. As such, we utilized time-dependent CLBR curves to trace the extent of CLBR improvements in each POSEIDON subgroup.
Cumulative live birth rates within 5 years
The cumulative incidence curves and pairwise log-rank comparisons for CLBRs between the optimistic and conservative methods in these five groups are shown in Figure 2. All curves differed significantly among these five groups (P<0.001). Under both the optimistic and conservative analytical methods, patients in all groups exhibited rapid increases in CLBRs within the first 1-2 years, with the CLBR of group 5 (control group) being the highest (optimistic: 76.2%; conservative: 70.5%), followed by that of group 1 (younger unexpected poor responders, optimistic: 75.6%; conservative: 63.7%), group 3 (younger expected poor responders, optimistic: 67.2%; conservative: 55.3%), group 2 (older unexpected poor responders, optimistic: 55.5%; conservative: 42.1%), and group 4 (older expected poor responders, optimistic: 35.7%; conservative: 25.9%).
Over the following 2- to 3.5-year periods, CLBRs in these five groups remained in the same rank-order and rose modestly when analyzed via the optimistic method (control group: 2.6%; group 1: 4.5%; group 3: 7.3%; group 2: 4.6%; group 4: 3.2%), whereas they rose more slowly when assessed via the conservative method (control group: 1.6%; group 1: 1.9%; group 3: 2.8%; group 2: 1.4%, group 4: 0.9%). Within 3.5-5 years, CLBRs in all five groups largely plateaued (control group: 0.5%; group 1: 0.7%; group 3: 0.5%; group 2: 0.2%, group 4: 2.5% with the optimistic method; control group: 0.4%; group 1: 0.4%; group 3: 0.3%; group 2: 0.1%, group 4: 0.4% with the conservative method).
Notably, among these five groups the CLBRs of patients in group 3 (younger expected poor responders increased most significantly within 2-3.5 years under both the optimistic (7.3%) and conservative (2.8%) analytical methods. Interestingly, the CLBRs of patients in group 4 (older expected poor responders) also rose notably within the atter 3.5-5-year period (optimistic: 2.5%; conservative: 0.4%), whereas the other four groups had largely plateaued by this time point. As the interval between oocyte retrievals or FET cycles differed among patients, this may not accurately reflect upward trends in CLBRs over time among groups. As such, we conducted further analyses of CLBR improvements as a function of the number of FET cycles completed in each of these POSEIDON subgroups.
Cumulative live birth rates within 9 FET cycles
Cumulative ongoing birth rates as a function of the number of FET cycles completed, as analyzed via both the conservative and optimistic approaches, are shown in Figure 3. CLBR curves differed significantly among groups with the exception of those for groups 1 and 5 (P<0.001). Under both the optimistic and conservative models, patients in these five groups exhibited rapid initial increases in CLBRs within 3 FET cycles, with these CLBRs decreasing from group 5 (control group, optimistic: 83.0%; conservative: 76.2%) to group 1 (younger unexpected poor responders, optimistic: 81.2%; conservative: 67.8%), group 3 (younger expected poor responders, optimistic: 77.1%; conservative: 60.7%), group 2 (older unexpected poor responders, optimistic: 60.5%; conservative: 45.5%), to group 4 (older expected poor responders, optimistic: 38.8%; conservative: 28.2%).
From 4-6 FET cycles, CLBRs in these five groups remained in the same rank-order as above and exhibited modest increases under the optimistic method (control group: 10.7%; group 1: 12.1%; group 3: 10.3% (only 5 FET cycles); group 2: 12.6%; group 4: 16.3%). While in the conservative method, the increase of CLBRs within five groups were extremely slowly (control group: 4.2%; group 1: 2.2%; group 3: 1.4%; group 2: 1.2%; group 4: 1.1%). Then CLBRs of five groups all maintained in their plateaus after 6 FET cycles, no matter in the optimistic or conservative methods.