After ten days of adaptive feeding, seventy-four specific-pathogen-free grade healthy female Sprague-Dawley rats, aged 8-week-old and weighing 180-200g (Animal License: SCXK (Beijing) 2011-0012), were involved in this study. These rats were housed in groups of 3 per cage and maintained under standard laboratory conditions (20-24°C; relative humidity: 50%-70%, light cycles: 10h light, 14h dark; bedding material kept dry).

These rats were then randomly divided into the Control group (n=8) and the POI model group (n=66), the latter intraperitoneally administered with 2ml of Cisplatin Injection (Jiangsu Hansoh Pharmaceutical Co. Ltd., lot number: H20040812, strength: 6ml:30mg) at a dose of 6mg/kg for once a week for two weeks to establish the model (9–13). The POI model group was then randomly divided into four groups: the model Control group (mo Control), the group treated with E2 (E₂), and the groups treated with low and high doses of Kuntai capsule (KTC-L, KTC-H).

For the 28-day administration, the Control and mo Control groups were intragastrically administered with 2.0 mL of 0.9% sodium chloride daily; the E₂ group with 2.0 mL of Estradiol suspension at a dose of 0.2mg/kg/d (Abbott Biologicals BV lot number:356933, Strength: 2mg/tablet); and the KTC-L and KTC-H groups with 2.0 mL of Kuntai capsule suspension (Guiyang Xintian Pharmaceutical Co. Ltd. lot number: 180405, Strength: 0.5g; 0.6g/kg/d, 1.8g/kg/d, respectively). Animal drug dose conversion was calculated based on the coefficient of equivalent dose transferring from rats to humans using body surface area (14) ( Figure 1 and Table 1 ).

The Shapiro-Wilk test was used to check if the continuous variables follow a normal distribution, and Levene’s test was used to assess the equality of variances. Variables normally distributed were shown as ‘mean ± SD’. One-way analysis of variance (ANOVA)-LSD was used for comparing group means. P < 0.05 was considered statistically significant.

The typical pictures of smears of vaginal cells at different stages of the estrous cycle in rats are shown in Figure 2A . The estrous cycles of the Control (5.18 ± 1.54) and KTC-L groups (5.46 ± 1.48) were significantly lower than the mo Control (8.35 ± 0.60) and KTC-H groups (7.2 ± 3.83) (P=0.000, P=0.006; P=0.000, P=0.02; respectively). No statistical significance was observed when comparing the KTC-L group with the Control and E₂ groups (P=0.676; 5.30 ± 1.83, P=0.809, respectively) ( Figure 2B ).

As shown in Figure 3A , the AMH and E₂ levels of the POI model group were significantly lower than the Control group (4.15 ± 0.92 vs. 3.34 ± 0.42, P=0.044; 378.12 ± 142.77 vs. 129.19 ± 33.31, P=0.005), while the FSH levels of the POI model group were significantly higher (17.64 ± 3.74 vs. 10.43 ± 2.09, P=0.019), thus indicating the successful establishment of cisplatin-induced POI rat model.

As shown in Figure 3B , the E₂ level of the Control group (172.69 ± 23.39) was significantly higher than the mo Control, E₂, and KTC-H groups (120.12 ± 25.92, P=0.028; 129.76 ± 25.37, P=0.035; 116.36 ± 19.90, P=0.011, respectively). However, the E₂ level of the KTC-L group (136.77 ± 38.78) was not significantly different from the Control group (P=0.082). The FSH level of the Control group (13.10 ± 2.32) was significantly lower than the mo Control group (18.65 ± 1.31, P=0.038) but not significantly different from the E₂, KTC-L, and KTC-H groups (16.05 ± 4.72, P=0.185; 15.30 ± 2.12, P=0.370; 15.46 ± 3.87, P=0.295; respectively). The AMH level of the Control group (4.18 ± 0.22) was significantly higher than the mo Control, E₂, and KTC-L groups (3.78 ± 0.27, 3.70 ± 0.36, 3.78 ± 0.34) (P=0.006, P=0.001, P=0.009), but not significantly different from the KTC-H group (4.02 ± 0.19, P=0.228). The AMH level of the KTC-H group was significantly higher than the E₂ group (P=0.020).

As shown in Figure 4 , the ovarian coefficient of the mo Control group (0.035 ± 0.004) was significantly lower than the Control (0.045 ± 0.007), KTC-L (0.045 ± 0.004), and KTC-H groups (0.044 ± 0.008) (P=0.003, P=0.003, P=0.009, respectively), with no statistical differences observed among the latter three groups. The E₂ group (0.040 ± 0.004) was not significantly different from the KTC-L and KTC-H groups (P=0.087, P=0.196, respectively).

As shown in Figure 5A , regarding the resting follicles numbers, the Control group (15 ± 8) were significantly higher than the mo Control, E₂, KTC-L, and KTC-H groups (3 ± 3, P=0.000; 8 ± 4, P=0.004; 6 ± 4, P=0.001; 7 ± 5, P=0.000; respectively). The E₂, and KTC-H groups were significantly higher than the mo Control group (P=0.012, P=0.032, respectively).

Regarding the numbers of the growing follicles, the mo Control group (6 ± 4) was significantly lower than the E₂, and KTC-H groups (19 ± 12, P=0.002; 15 ± 9, P=0.017; respectively). Figure 5B is the typical HE pictures of follicles of the five groups.

As shown in Figure 6 , regarding the relative protein expression of GDF9, both the KTC-L (0.985 ± 0.188) and KTC-H groups (0.705 ± 0.038) were significantly higher than the Control (0.530 ± 0.061) and mo Control groups (0.551 ± 0.035) (P=0.000, P=0.000; P=0.013, P=0.027, respectively).

About the relative protein expression of LC3A-II, the KTC-H group (1.080 ± 0.424) was significantly higher than the Control, mo Control, and E₂ groups (0.641 ± 0.010, P=0.001; 0.746 ± 0.035, P=0.011; 0.780 ± 0.013, P=0.021; respectively). The KTC-L group (0.988 ± 0.201) was significantly higher than the Control group (P=0.009).

About the relative protein expression of Beclin 1, both the KTC-H (0.832 ± 0.014) and KTC-L groups (0.843 ± 0.088) were significantly higher than the Control (0.711 ± 0.110) and E₂ groups (0.688 ± 0.011) (P=0.004, P=0.001; P=0.002, P=0.000; respectively). The mo Control group (0.781 ± 0.044) was significantly higher than the E₂ group (P=0.023).

GDF9 is a member of the Transforming Growth Factor β (TGFβ) family. In addition to promoting pre-antral follicles growth, antral follicles growth, and granulosa cells proliferation, GDF9 also stimulates the granulosa cells’ basal secretion of progesterone and Estradiol of small antral follicles of female rats (16–19). In this study, GDF9 expression in both Kuntai groups was significantly higher than in the Control and mo Control groups. This indicates that Kuntai capsule has a positive effect on the development of granulosa cells in small follicles, which is consistent with the results of follicle counts.

Microtubule-associated protein 1A/1B-LC3 is a soluble protein widely distributed in mammalian tissues and cultured cells. The expression of LC3A-II in the mammalian ovaries is relatively higher than other isoforms, thus making it a suitable marker for monitoring autophagy (20–24).

The results of this study revealed the expression of LC3A-II protein in the following descending order: KTC-H > KTC-L > E₂ > mo Control > Control. In other words, all four POI model groups had a higher LC3A-II expression than the Control group, with the KTC-H and KTC-L groups showing the most apparent autophagic responses.

The Beclin 1 gene is the mammalian ortholog of yeast autophagy-related gene-6 (ATG6) and the most widely studied autophagy-related gene. Beclin 1 acts as a platform for binding with proteins like Phosphoinositide 3-kinase (PI3KC3), ultraviolet radiation resistance-associated gene (UVRAG), activating molecule in Beclin 1-regulated autophagy (Ambra1), and Bax interacting factor-1 (Bif-1) to promote autophagy (25–29).

In this study, the expression of Beclin-1 is in the following descending order: KTC-L > KTC-H > mo Control > Control > E₂. The results showed that although autophagy is common among the five groups, Beclin-1 expression in both the KTC-L and KTC-H groups was significantly higher, indicating that Kuntai Capsule induces stronger autophagy than the other administration.

Autophagy plays dual roles in cell survival and death by inhibiting or partnering with apoptosis. Autophagy helps the body eliminate cancerous cells, and damaged organelles facilitate cellular adaptation to the changing internal and external environment and promote cell survival by delaying aging and cell death (30, 31).

Unlike modern medicine, which has a single active compound, Traditional Chinese Medicine integrates many plants into formulae of compatible ingredients. Each ingredient contains various compounds, with complex interactions among these compounds forming a network. This also gave rise to the development of the discipline ‘Traditional Chinese Medicine Network Pharmacology’ (32–35), which updates the research paradigm from the current ‘one target, one drug’ mode to a new ‘network target, multi-components mode’ (36).

Research on the network pharmacology of Kuntai Capsule has been carried out recently. Kuntai Capsule consists of 6 herb components: Rehmannia glutinosa, Coptis, Paeonia, Scutellaria, Equus asinus, and Poria cocos. Li DX (37) utilized the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) to find out 21 active compounds corresponding to the six herb components, which are mainly flavonoids and alkaloids, including verbascoside, 5-hydroxymethylfurfural, baicalein, wogonin, skullcapflavone II, berberine, columbamine, magnoflorine, jatrorrhizine, and palmatine, etc. Wang LJ and Chen YM (38, 39) did further research and acquired the corresponding 186 gene targets in the cells of the human body of these 21 compounds via TCM-mesh (http://mesh.tcm.microbioinformatics.org) and COSMIC (http://cancer.sanger.ac.uk/cosmic). These gene targets are highly enriched in the central nervous system, breast, endometrium, and ovary. This result indicates that the Kuntai capsule can treat related diseases caused by ovarian insufficiency. Besides, Kyoto Encyclopedia of Genes and Genomes (KEGG) results found that these targets were involved in the PI3K-Akt, mTOR, and insulin signaling pathways, which participate in follicular development in the follicle recruitment stage (38–40). Signaling pathways may become one of the directions in further research.

Several studies used different doses of cisplatin to establish the POI rat model, the median lethal dose (LD₅₀) of cisplatin in rats ranging from 6 to 20 mg/kg (41–43). A dose less than 6 mg/kg can reduce death, but the POI model would not be successful enough to carry out further experiments; doses higher than 6 mg/kg would ensure a higher success of modeling, but it would have adverse effects such as decreased appetite, weight loss, and myelosuppression, and especially a higher risk of death (9). In a study (10) comparing lower doses of cisplatin (1.5 mg/kg, 2.0 mg/kg, or 2.5 mg/kg once a day for five consecutive days) with higher dose of cisplatin (4 mg/kg on Day 1 and 6 mg/kg on Day 8), it was concluded that the lower dose regimes were not effective to establish the POI model by examining the body mass, follicle count, and serum AMH.Therefore, in our pre-experiment study, we did not adopt the lower-dose cisplatin method to establish the POI model and followed the higher-dose modeling method (9).

Our mortality rate, at 41%, was about in the range of other studies as far as mortality rates have been reported (41–43). Contrary to those studies that do not mention their mortality rate in the modeling process, we intended to list out the number of rats that died in the modeling process to act as a reference for researchers planning to use cisplatin-induced POI rat models and a reminder that other means need to be explored to reduce mortality rate while ensuring the modeling effectiveness, such as caudal injection of cisplatin or subcutaneous injection of saline to reduce dehydration. Our intention was to stress the need and the possibilities to reduce the mortality due to the toxic cisplatin.