Ovulatory dysfunction, a condition with multifactorial causes, is one of the major causes of female infertility [1], accounting for over 50% of infertile etiologies [2]. In 1973, the World Health Organization (WHO) classified anovulatory patients into three groups based on the levels of gonadotropins and estrogens [3], namely, WHO group I: hypogonadotropic hypogonadal anovulation, WHO group II: normogonadotropic anovulation and WHO group III: hypergonadotropic hypogonadal anovulation. The most problematic and common anovulation is WHO group II which includes women with hypothalamic-pituitary-ovarian dysfunction. Cases diagnosed with polycystic ovarian syndrome (PCOS) are the most prominent in WHO group II [3, 4]. The classification has been used widely as a guidance for ovulation induction [3].

Clomiphene citrate (CC), a triphenylethylene derivative has traditionally been used as the first-line ovulation induction therapy for anovulatory women [5, 6, 7, 8]. As a selective estrogen-receptor modulator, clomiphene citrate reduces sex steroidal negative feedback at the hypothalamic-pituitary axis, thereby enhancing secretion of follicle-stimulating hormone (FSH) and luteinizing hormone (LH) from the anterior pituitary gland. This action, which stimulates the growth of the ovarian follicle and selection of dominant follicle and thus initiates ovulation, is initiated within the first five days of menstrual cycle under the theorical basis that physiologic lower levels of FSH permit dominant follicle selection [8]. The starting dose is usually 50 mg per day, as the ovulation rate is generally between 50% and 75% in PCOS cases [9]. Among responders, successful ovulation, singleton and multiple pregnancy was reported at 75%, 15–47% and 8–10%, respectively, within six months of ovulation induction [6, 8, 10]. Some studies reported a 74% chance of ovulation with a starting dose of CC at 100 mg/day [8]. Evidence suggests that CC was 1.35 to 20 times more likely to trigger ovulation compared to placebo [5].

Responsiveness to CC for each individual depends on numerous factors, including age, body weight and biochemical blood test [11]. In our clinic, CC is the first-line drug prescribed for ovulation induction in anovulatory women in either artificial insemination cycles or timing for sexual intercourse cycles. Overall ovulatory rate was 75–80%. Previous studies demonstrated that younger age, lower BMI, less insulin resistance and less hyperandrogenism were all found to be predictive of ovulation [12, 13, 14]. Our study aimed to investigate the predicting factors of responsiveness to CC in infertile women with WHO group II among the Thai population. Accordingly, the enrolled participants were a broader subset than PCOS patients. Secondary outcome included prediction of ovulation success rate and further appropriate treatment planning via CC. Alternative treatment options for the predicted non-responder were also recorded.

A retrospective case control study was performed at Thammasat Fertility Center of Thammasat University Hospital, Thailand. A total of 260 patients meeting the criteria were enrolled between January 2017 and August 2020.

The sample size was calculated from G*power 3.1.9.4 for case control study [15]. The case group was responders to CC, and control group was non-responders, with an allocation ratio of 2:1. The effect size was set to 0.5 based on the study conducted by Sachdeva G et al. [13] with power (1-$\beta$ error) at 0.95. The resulting sample size was 157 in the case group, and 79 in the control group. In order to prevent loss of information along the process, the population was increased by at least 10% in each group.

The CC responsive group (CCR) was defined as patients who were examined by transvaginal ultrasound with at least 1 dominant follicle $\ge$17 mm or at least 1 dominant follicle $\ge$15 mm, provided that the attending physician concluded they should receive ongoing treatment within the cycle. The CC non-responsive group (NCCR) was defined as patients who did not meet the responder group criteria and whose cycles were cancelled, accordingly.

The participants included in this study were all anovulatory women WHO group II, 20–40 years of age who received CC in initial treatment at Thammasat Fertility Center of Thammasat University Hospital. The cases were considered as anovulatory WHO group II if they had baseline FSH between 4–10 mIU/mL on early menstrual days (day 2–3) combined with at least one of the following conditions:

(1) Had a history of irregular menstrual period longer than 35 days, or

(2) Presented with oligomenorrhea, or

(3) Presented with secondary amenorrhea, or

(4) Reported no ovulation with the use of at least three cycles of self LH tests.

Excluded from the study were participants with a body mass index (BMI) lower than 18.5 or higher than 35 kg/m${}^2$, a history of ovarian or fallopian tubal surgery, history of endocrine disorder affecting ovulation (hyperprolactinemia, thyroid, and adrenal gland diseases), prior use of metabolic disease medication (insulin sensitizing drugs, lipid-lowering drugs, anti-obesity drugs, gonadotropin-releasing hormone analogs, oral contraceptive pills, anti-androgenic drugs, ovulation-inducing agent, dopamine agonist drugs), and those with incomplete data. Flow chart of the study is shown in Fig. 1.

Fig. 1.

Flow chart of study. Recruitment: infertile women with WHO group II anovulation who received clomiphene citrate 50 mg 2 tabs oral at day 2–5 of cycle, CCR, CC responsive group; NCCR, non-CC responsive group, study group: patients who were examined by transvaginal ultrasound to record at least 1 dominant follicle $\ge$17 mm; control group: patients who did not meet the study group criteria.

All participants received standard care, including medical history on obstetric-gynecologic history, infertility history, and general physical examination, including body weight, height, waist and hip circumference measurement. The waist circumference was measured midway between the lower rib margin and the iliac crest in the mid-axillary line at the end of normal expiration. The hip circumference was measured at the highest prominence of the buttocks parallel to the floor [16, 17]. Patients underwent transvaginal ultrasonography (TVS) for pelvic organ evaluation, measurement of antral follicles in early follicular phase (day 1–5 of cycle) and baseline hormonal measurement, including follicle stimulating hormones, luteinizing hormones, prolactins and estradiol levels. Low antral follicle count was defined as AFC $\le$5 follicles. Participants who agreed to start the ovulation induction cycles were treated with CC 100 mg/day (2 tablets of CC 50 mg) oral at bedtime, starting within second to fifth day of cycle for 5 consecutive days. Treatment was followed by TVS at day 10–13 of cycle to check for follicular growth. The TVS was performed by five certified reproductive medicine specialists. If dominant ovarian follicular size did not meet the responding criteria, patients were appointed for the next ultrasonographic evaluation in 2–7 days, but not exceeding the 28th day.

Data were collected from the medical records of fertility clinic. The acquired information included age, body weight, height, waist circumference, hip circumference, obstetric data, menstrual cycle, diagnosis of PCOS, lifestyle habit (such as alcohol drinking and smoking), current medication and baseline hormones. Information concerning number of visits as well as follicular growth and size was also recorded.

Statistical analysis was done by SPSS Statistics Version 23 (IBM Corp., Armonk, NY, USA). The continuous parameters were recorded as mean and standard deviation (SD). Normality of quantitative data was calculated by Kolmogorov Smirnov tests. Independent t-test or Mann Whitney U test was used for two independent variables based on whether the data was normally distributed. Frequency and percentage were used to represent category parameters. Medium effect size, alpha value (two-sides) and power significant level were set at 0.5, 0.05 and 0.95, respectively. Logistic regression analysis was used for factors associated with the CC responsiveness.

A total of 260 infertile women with normogonadotropic anovulation (WHO group II anovulation) who received CC as an ovulation induction were enrolled in this study.

Study (CC responder: CCR) and control (non-CC responder: NCCR) groups consisted of 173 and 87 cases, respectively. Demographic characters among study CCR group and NCCR group were compared as represented in Table 1. The only difference of statical significance was that waist-hip ratio (WHR) of NCCR was greater than that of CCR group.

Regarding anthropometric measures, the average body mass indexes were 23.9 $\pm$ 10.7 and 24.0 $\pm$ 4.0 kg/m${}^2$. There were 77 (44.51%) participants in the responsive group and 48 (55.17%) in non-responsive group who had obesity (BMI $\ge$23 kg/m${}^2$). No significant differences between the groups were detected. Interestingly, WHR were 0.81 $\pm$ 0.05 and 0.83 $\pm$ 0.06, respectively (p = 0.004).

Regarding ovarian reserves, baseline hormonal levels (FSH, LH, prolactin and estradiol) of infertile cases were comparable between both groups as cited in Table 2. These biological hormones were at normal range in early follicular phase of reproductive age. The number of participants who had low antral follicle counts (AFC $\le$5 follicles) was 30 in CCR (17.34%) and 12 (13.79%) in NCCR group. Hyperandrogenism state among participants was 57.3% (149/260).

Regarding clinical characteristics of PCOS, the patients who presented with at least one of the clinical characteristics of hyperandrogenism, such as hirsutism, oily face skin, and acne were 99 (57.23%) and 50 (57.47%), a difference that was not statistically significant. Participants with PCOM on TVS were 104 (60.12%) and 58 (66.67%), respectively.

Subgroup analysis of ovulatory group is shown in Table 3. A total of 173 participants were divided into two groups, single follicular development and multiple follicular development. The average age of multifollicular developing patients was significantly lower than that of patients with single follicular development, i.e., 31.7 $\pm$ 4.1 and 33.0 $\pm$ 3.8 years, respectively (p = 0.037). Other parameters were not statistically different.

In univariate logistic regression analysis between CCR and NCCR, WHR was the only factor associated with successful ovulation prediction (odds ratio 1.97, p = 0.02). After performing binary logistic regression analysis with the BMI group, the adjusted odds ratio was 1.789 (p = 0.06), a potential predictor of CC-responsiveness.

Receiver operating curve (ROC) was generated to predict the CC responsiveness as shown in Fig. 2. The best cut point value of WHR at 0.775 was chosen to give the appropriate area under curve (AUC). At this cut point, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were 90.8, 20.2%, 44.9% and 70.7%, respectively.

Fig. 2.

Receiver operating curve (ROC) of waist-hip ratio for predicting clomiphene responsiveness.

WHO group II anovulation is the most common anovulatory infertility worldwide. The majority of cases (80%) are polycystic ovary syndrome (PCOS) which is usually associated with underlying metabolic abnormalities. Patients diagnosed with PCOS are often obese and have insulin resistance associated with resistance to ovulation, lower pregnancy rate, live birth rate and obstetric complications [18, 19].

Clomiphene citrate has been widely accepted as the first line ovulation induction medical treatment [19], regarding its high efficacy, easy to administer, safety profile, and low cost. However, the variation of clomiphene responsiveness has been reported [10, 11, 12, 13].

Kuang H. et al. [12] validated the predictive model for ovulatory and pregnancy outcome for PCOS patients based on the cumulative study of 1376 women in 2015. The ovulatory drugs in the study compared clomiphene citrate (CC) to metformin or their combination (PPCOS-I) and clomiphene citrate compared to letrozole (PPCOS-II). Younger age, lower BMI, shorter duration of attempting to conceive, less insulin resistance and less hyperandrogenism were all found to be predictive of ovulation, conception, and clinical pregnancy. Nevertheless, there was a minor discrepancy between the PCOS I and II.

Our study findings support the evidence that patients with higher WHR, especially those with truncal obesity, are less likely to respond to CC [20]. After performing logistic regression analysis, we found that truncal obesity remained a significant predictor of ovulation.

Sachdeva G. et al. [13] conducted a prospective observational study in infertile PCOS women who received incremental doses of clomiphene citrate from 50 mg/d to a maximal dose 150 mg/d. The results demonstrated that BMI and WHR could be used as the anthropometric predictors for clomiphene responsiveness. The study concluded that BMI is the best predictor. Nevertheless, BMI was not significantly different between the two groups in our study. This can be explained by the following reasons, Firstly, the mean BMI among our participants is lower than that in previous reports, and did not reflect the common characteristics in those populations. Secondly, underscoring a difference of interventions, our study recruited only the first cycle of participants to examine the effect of 100 mg clomiphene in a single cycle, since only 11.7–13% of participants who did not respond to CC 50–100 mg per day could ovulate with CC 150 mg per day [14, 21]. Therefore, the NCCR group in our study was not comparable to “clomiphene resistance”. Consequently, this might limit the power of BMI to detect any significant difference. Lastly, the underlying pathophysiology causing the anovulatory failure in PCOS is well defined. Truncal obesity is more associated with visceral fat distribution, insulin resistance and ovulatory failure than obesity diagnosed by BMI [22, 23]. As such, truncal obesity is more sensitive than BMI in predicting clomiphene responsiveness.

As for the clinical feasibility of WHR to predict CC responsiveness, we therefore used the ROC curve. The cut-off value to predict drug responsiveness is 0.775, with 90.8% sensitivity and 20.2% specificity. This value could be used for considering other treatment strategies. There are many proposed strategies for improving ovulatory outcome. The first strategy entails intervention to reduce truncal obesity such as preconception lifestyle modification, including weight loss program [24, 25], caloric restriction, behavioral modification and increased physical activity [26]. These can improve both reproductive, metabolic outcome and ovulation with clomiphene citrate [27]. The second strategy involves other ovulation-stimulating agents such as letrozole, clomiphene combined with other agents such as metformin, gonadotropins or letrozole [18, 27, 28, 29]. Letrozole, the first-line ovarian stimulating agent, was found to be more effective than clomiphene in PCOS patients [11, 30]. Letrozole significantly increases ovulation, pregnancy and live birth rate [4, 11, 19]. The combination of letrozole with clomiphene citrate was also reported in some studies. The WHR cut-off value supports our clinical practice guideline for ovulation induction and can also be used in counseling sessions.

The current study did not include participants who were taking metformin. The data showed that clomiphene citrate combined with metformin resulted in statistically significant higher ovulation rates [4, 31]. Hashim et al. [32] conducted a meta-analysis in 2015 and found that the ovulatory rate by clomiphene citrate plus metformin was higher than clomiphene alone (odds ratio 1.55, 95% CI 1.02–2.36) in the PCOS population with insulin resistance. Wang R. et al. [5] found that the clinical pregnancy rate was greater in clomiphene plus metformin group compared to the clomiphene group alone (8 RCTs, 1039 women, RR 1.18, 95% CI 1.00–1.39).

A limitation of our study is that we recruited the group of normogonadotropic anovulatory women including PCOS who share distinct pathophysiology from the rest of this group. Our study demonstrated that PCOS is not a predictive factor for clomiphene resistance. But we should interpret the data carefully. In our center, not all participants were evaluated by the serum androgen levels. Therefore, the accurate prevalence of PCOS in our participants is still uncertain and may be underestimated. The outcome is only “ovulation” which is the surrogate marker for fertility treatment. Also, the endocrinologic data was limited due to the nature of retrospective review. Further prospective study with androgen and metabolic profile with a focus on pregnancy outcomes or a well-designed, randomized controlled trial with other agents in this population are recommended.

Our study demonstrated that waist-hip ratio greater than 0.775 is associated with non-responsiveness to clomiphene citrate in ovulation induction cycle and can be considered as a predictor for clomiphene responsiveness in normogonadotropic, normogonadal anovulatory patients. Alternative ovulatory drugs or interventions to reduce WHR should be considered in such cases to maximize the ovulatory, and hence fecundity, outcomes.

AFC, antral follicle count, BMI, body mass index; CC, clomiphene citrate; CCR, Clomiphene responsive group or ovulatory group; NCCR, Clomiphene non-responsive group or anovulatory group; FSH, follicle-stimulating hormone; LH, luteinizing hormone; PCOM, polycystic ovarian morphology; PCOS, polycystic ovarian syndrome; WHO, world health organization; WHR, waist-hip ratio.

TA and AS conceived and designed the study. All authors took parts in data collection and evaluation, drafting and statistical analysis. TA performed data collection, statistical analysis, and interpretation of data under supervision by JP and AS. TA Drafted the manuscript, which was revised by AS and KS. All authors performed editing and finalization of the manuscript. All authors read and approved the final manuscript.

This study was approved by Human Research Ethics Committee of Thammasat University No.1 (MTU-EC-OB-1-162/62). This study was designed as a retrospective assessment of data, and therefore, informed consent was not required.

The authors would like to express their gratitude to all the doctors, scientists and medical staffs in Thammasat fertility center, department of Obstetrics and Gynecology, Thammasat university hospital.

This trial was supported by Faculty of Medicine, Thammasat University research fund No.1-08/2563.

The authors declare no conflict of interest.