Prediction of Patients with Ovarian Hyperstimulation Syndrome in Gnrh-A Prolonged Protocol : A Retrospective Study

Background: OHSS (ovarian hyperstimulation syndrome) is a life-threatening complication and most common adverse effect of fertility treatment Objectives: To investigate associated risk factors of ovarian hyperstimulation syndrome (OHSS) in stimulated ovarian cycles with assisted reproductive technology, may provide guidance for clinicians on how to prevent OHSS. Methods: A Logistic regression analysis was conducted in 336 patients who had completed IVF-ICSI/ET cycles between April 2019 and April 2020 in the rst aliated hospital of kunming medical university. Receiver-operating characteristic (ROC) curves for OHSS were calculated for each predictor using cut-off values. Area under the curve (AUC) analysis and logistic regression models were performed to compare the performance of laboratory biomarkers. Results: According to clinical diagnosis, 61 cycles developed OHSS of 336 cycles, with an incidence of 18.15%. Patients were graded according to their clinical symptoms and severity, including 27 cycles of moderate OHSS (8.04%) and 8 cycles of severe OHSS (2.38%). The cut-off value of AMH, E2 on HCG day, egg retrieved to predict moderate and severe OHSS were 7.495ng/ml, 4828pg/ml and 19.5 in GnRH-a Prolonged Protocol, with sensitivities of 77.0%, 67.2%, 80.3%, and specicity of 68.0%, 70.9%, 81.5% respectively. The area under the curve (AUC) values of AMH, E2 on HCG day, Number of oocytes retrieved to predict moderate and severe OHSS were 0.754, 0.738, 0.876, respectively. And the AUC value of combined index to predict moderate and severe OHSS was 0.898, achieved the highest AUC with 85.2% sensitivity and 83.6% specicity. Conclusions:AMH, E2 day of HCG and number of eggs obtained have a good predictive effect on the occurrence of OHSS, and the predictive ability is further improved after the combination of the three indicators.

Introduction ovarian hyperstimulation syndrome (OHSS) is a serious life-threatening iatrogenic complication, it may occur spontaneously during pregnancy. The main pathophysiological mechanism is the increase of capillary permeability [1]. HCG is considered to be the main triggering factor of the syndrome [2]. Increased vascular permeability leads to uid transfer from intravascular space to extracellular space. Women with high estradiol level, large number of follicles or oocytes retrieved, and polycystic ovary syndrome (PCOS) patients seem to have a particularly high risk of developing OHSS [3]. However, early identi cation of potential risk factors of OHSS and clinical intervention can signi cantly reduce the incidence of OHSS. Therefore, early prediction of OHSS occurrence and timely intervention measures are important links to improve the safety of assisted reproductive technique (ART), With the increasing use of ART, fertility physicians need to be equipped with the knowledge to control the occurrence of OHSS. But the indicators that predicted the occurrence of OHSS were different, some studies suggest that AFC and basal E2 are Page 3/12 strong predictors of OHSS [4], other risk factors of OHSS include age, low body mass index (BMI), polycystic ovary syndrome (PCOS), and previous OHSS history, serum anti-Müllerian hormone (AMH), antral follicle count (AFC), estradiol (E 2 ) on the day of human chorionic gonadotropin (hCG) injection and the number of oocytes retrieved were also predictors of OHSS [5]. A single indicator has limitations in the sensitivity and speci city of OHSS prediction, but the combination of multiple indicators may improve the prediction of disease [6,7], however, there is no study on the prediction of OHSS with multiple indicators.
GnRH-a Prolonged Protocol was considered have higher cumulative live birth rate in POSEIDON group 4 patients with higher anti-Müllerian hormone levels (anti-Müllerian hormone ≥ 0.785 ng/mL) than gonadotropin-releasing hormone antagonist protocol[8], and has higher CLBR, normal fertilization rate and number of available embryo than GnRH-ant group in a RCT research including 18853 women from China [9],but didn't increase the risk of OHSS compared to GnRH antagonist protocol [10]. However, at present, there is no study on the risk factors of OHSS in this regimen.  [12].

Measurements and laboratory analyses
Gonadal hormone concentrations and AMH measurements were performed on the serum of each subjects. Level of AMH will be examined using a Gen 2 ELISA kit (Beckman Coulter, Inc. Brea, CA, USA). Serum was stored at −70℃. A combined gas chromatographic negative ionization tandem mass spectrometry and liquid chromatographic electrospray tandem mass spectrometry bioanalytical method was used to measure Gonadal hormone (Taylor Technology, Princeton NJ).

Early-Follicular Phase Gonadotropin-Releasing Hormone Agonist (GnRH-a) Prolonged Protocol
One depot of 3.75 mg GnRH-a was injected on day 2 of the menstrual cycle, 35 days later, pituitary downregulation was confrmed (no follicular diameter >8 mm, E2<50 pg/ml), Gonadotropin (rFSH-Gonal-f Merck Serono, Germany) was used at an initiative dose ranging from 150-225 U. The dose of Gn was adjusted according to the size, counts of follicles and hormone levels. When 2 dominant follicles reached 18 mm in diameter, or three follicles = >17 mm, human chorionic gonadotrophin (HCG, 5000-10,000 IU, Lizhu Pharmaceutical Trading Co., German) was administered for trigger ovulation [13].
If number of oocytes retrieved ≥15, the serum E2 ≥ 5000 pg/ml on hCG trigger day or the serum E2 ≥ 1500 pg/ml on the second day after oocyte retrieval, or OHSS manifestations such as nausea, vomiting, abdominal pain and distension appeared, patients would receive a "freeze-all" approach instead of fresh embryo transfer to prevent late OHSS [12].
There were 61 cases of OHSS and 275 cases of non-OHSS. The database was collected after the approval of the medical ethics committee. All data were collected from the electronic medical record system and adjusted by at least two authors.
All of the included patients read and signed the informed consent form. This retrospective study was approved by the Ethics Committee of the First A liated Hospital of Kunming Medical University. All the treatments in the present study were performed strictly in accordance with the Declaration of Helsinki for Medical Research.

Statistical analysis
All analyses were performed using SPSS 22.0 statistical software package (SPSS Inc, Chicago, IL).
According to the distribution of data types to choose the appropriate statistical methods to analyze the differences between groups, two independent groups were assessed using Student's t-test or the Wilcoxon rank-sum test. We constructed a multivariable logistic regression model by including each of the conditions as independent variables. The dependent outcome variable was OHSS or not. Receiver operating characteristic (ROC) analysis was carried out and area under the curve (AUC) was compared to evaluate the predicting ability of clinical indicators scores. Based on the cut-off values, sensitivity and speci city were also calculated. A Z test was used for comparing the AUCs between different curves. The Bonferroni method was used adjust for multiple comparisons.

Results
Comparison baseline characteristics between the OHSS and non-OHSS groups A total of 336 cycles were included in this study. According to clinical diagnosis, 61 cycles developed OHSS, with an incidence of 18.15%. Patients were graded according to their clinical symptoms and severity, including 27 cycles of moderate OHSS (8.04%) and 8 cycles of severe OHSS (2.38%).
Basal FSH of the OHSS group was lower than the non-OHSS group (p <0.001). The values of basal LH, basal T, number of AFC, AMH and TSH in the OHSS group were all higher than those in the non-OHSS group, the proportion of PCOS patients in OHSS group was higher than that in control group (p <0.001).
There was no signi cant difference in age, BMI, the duration of infertility, BMI, basal E2 value between these two groups (p>0.05, Table 1).

Clinical parameters between patients with and without OHSS
The Median of Gn at an initiative dose was about 150 IU in both non-OHSS and OHSS group. The E2 and P on HCG day in the OHSS group was higher than that in the non-OHSS group. The number of oocytes retrieved in the OHSS group was all higher than non-OHSS group(p<0.001). But the Median of total Gn dose in the OHSS group was lower than the Non-OHSS, patients with OHSS are more sensitive to Gn. After adjusting age, AMH and the number of sinus follicles, no statistical difference was found in the clinical pregnancy rate between the two groups (Model )( Table 2). None of the OHSS patients was Delayed OHSS.

Logistics regression analysis
Multiple logistic regression analysis was used to adjust for case-mix differences between OHSS and non-OHSS groups. Basic AMH, E2 on HCG day and the number of eggs obtained were found to be strongly predictive of OHSS (Table 3).

ROC curve
Cut-off value for predicting the occurrence of OHSS were determined as follows: AMH≥0.75ng/ mL, HCG The sensitivity of combination indicators was 83.6%, and the speci city was 85.2%, both of which were improved compared with the single index, suggesting that the combination of multiple indicators has a certain signi cance in predicting the occurrence of OHSS, according to the ROC curve, the area under the curve (AUC) after combining the four indicators is 0.898 (0.854-0.942, Table 4), which strengthens the predictive power and is higher than the AUC for any single indicator to predict the occurrence of OHSS (Fig1).

Discussion
OHSS is an iatrogenic complication caused by an overreaction to gonadotropin stimulation. OHSS is a common disease during controlled ovarian hyperstimulation treatment, some patients even need to seek initial care in the emergency department because of multiple complications of OHSS, including including ascites, hemodynamic instability, thromboembolic disease, renal insu ciency, acute respiratory distress syndrome, even GDM and neonatal NICU admission and so on [14]. It has been reported in the literature that moderate to severe OHSS accounts for 3-10% of all ART cycles, and the incidence is as high as 20% in high-risk women [15]. Although ART is considered safe, but moderate to severe overstimulation syndrome (OHSS) is considerd as a complication have signi cant morbidity and mortality [16]. Compared with the antagonist protocol, the GnRH-a Prolonged Protocol had higher cumulative live birth rate[8], and the incidence of OHSS was not different from that of the antagonist protocol [10]. Due to its high cumulative live birth rate, it is also a commonly used prescription in clinical practice at present. However, there is no study on the risk factors for OHSS in the GnRH-a protocol.
Our study found that the risk of OHSS in patients of GnRH-a Prolonged Protocol was similar to that of other protocol s, including high AMH value, high number of eggs retrieved, and high estrogen Besides,Young age, lean body mass, Polycystic ovary syndrome (PCOS), Gonadotrophin dosage and so on were consider as risk factors for OHSS [17]. This is the same with our study, younger age also observed in the OHSS group compared to non-OHSS group, but nally, the age factor was not included in the regression analysis, and age could not be used as an indicator to predict the occurrence of OHSS.
Because of the harmfulness of OHSS, looking for indicators to predict the occurrence of OHSS can prevent OHSS in advance. As Akanksha et.al reported [18], the optimal thresholds of AMH for predicting OHSS was 22.5 pmol/L, thresholds of AFC was 19.5, and egg numbers was 9.5, this is similar to our research, which AMH≥4.295ng/ mL, and egg numbers ≥18.5 can predict the occurrence of OHSS. But in their research, peak estradiol levels had no predictive value, this is different from what we found that HCG day E2≥4824pg/ mL also can strongly predict the occurrence of OHSS with a sensitivitie of 70.9%. Since the sensitivity of single indicator to predict OHSS was between 70-90%, and the speci city was between 59-88%, we used the combined indicator to predict OHSS, and found that the combined indicator had both good sensitivity (83.6%) and speci city (85.22%), which could better predict the occurrence of OHSS.
PCOS also is thought to be an important risk factor of OHSS during ovulation induction [19]. But whether the patient with PCOS can't predict the occurs of severe OHSS in our study, even though the proportion of PCOS patients was higher in the OHSS group, this may be due to the pretreatment we have done in patients with PCOS,such as metformin used in PCOS patient with insulin resistance, has also been studied can reduce the occurrence rate of OHSS[20]. In our study, the number of eggs obtained also can predict OHSS, but the number of eggs obtained is uncontrollable because steering the ovarian response into a supposed optimal range may appear di cult [21], so the number of eggs can't be a reference factor when we formulate the starting dose of Gn for patients, the same with the E2 value on HCG day.

Conclusion
Patients with serum AMH levels above 7.495ng/ mL, HCG day E2 levels above 4824pg/ mL and egg numbers ≥19.5 can predict OHSS in GnRH-a Prolonged Protocol, but combine these three indicators can predict OHSS better.  Data are presented as the median lower quartile upper quartile or count (percentage), *p<0.05

Declarations
Model : The occurrence of OHSS was the independent variable (0 = non-OHSS, 1 = OHSS), and clinical pregnancy was used as the dependent variable (0 = not pregnancy, 1 = pregnancy), Model on the basis of the model adjusted for age, Model On the basis of the model adjusted for Age, AMH, AFC.  Figure 1 Receiver-operating characteristics curve comparisons of different parameters in predicting OHSS.