This retrospective study included 360 patients with PCOS, diagnosed according to the Rotterdam criteria (15), who were treated at our hospital from January 2021 to January 2024. All participants signed informed consent forms. Patients were divided into three groups based on BMI: normal weight group (BMI 18.5–24.9), overweight group (BMI 25–29.9), and obese group (BMI > 30) (16). The exclusion criteria were as follows: (1) patients with other reproductive system diseases and endocrine disorders, such as thyroid dysfunction (hyperthyroidism or hypothyroidism), hyperprolactinemia, and Cushing’s syndrome; (2) pregnant or lactating women; (3) severe liver and kidney dysfunction; and (4) patients with mental or cognitive impairments. This study was approved by the Ethics Committee of Hubei Cancer Hospital, Tongji Medical College, and Huazhong University of Science and Technology and conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent. Using a medium effect size, a significance level of 0.05, and a statistical power of 80%, for continuous variables, the required sample size was at least 53 participants per group. For categorical variables, the total required sample size was 240, with an average of 80 participants per group. The sample size in this study meets these requirements. A total of 412 PCOS patients were initially enrolled in the study. After the intervention and follow-up, 360 patients completed the full intervention and follow-up, with 52 dropouts, corresponding to a dropout rate of approximately 12.6%. The primary reasons for dropout included loss to follow-up, incomplete adherence to the dietary plan, or personal reasons.

The nutritional intervention in this study consisted of comprehensive dietary counseling and a personalized nutrition plan. Each patient received guidance from a professional nutritionist, including the establishment of healthy eating habits, food selection, knowledge of nutritional components, and strategies to maintain a balanced diet in daily life. The intervention emphasized a reasonable distribution of macronutrients: protein accounted for approximately 15–20% of total energy, carbohydrates 40–55%, and fats 25–30%. For PCOS patients with normal weight, the total calorie intake is controlled at 1,200–1,500 kcal/day, while for overweight or obese patients, the calorie intake is 10–20% lower than that of normal patients. The protein intake was set at 1.0–1.2 g/kg/body weight per day and carbohydrate intake was prescribed at 225–325 g/day, with overweight patients consuming approximately 180–225 g/day and obese patients consuming approximately 120–160 g/day. The specific energy intake for each patient was individualized according to their BMR, ensuring that total caloric intake was slightly above BMR to meet basic metabolic needs. Carbohydrate intake was recommended to account for 40–55% of total energy requirements and was further adjusted based on patients’ clinical indicators (e.g., insulin resistance index, fasting plasma glucose, and lipid profile). This approach ensured nutritional balance while maximizing the effectiveness of the intervention. The intervention lasted for 3 months.

With regard to dietary quality, this study emphasized the rational selection of nutrient sources and food choices. Carbohydrates were primarily derived from low-glycemic index (GI) and high-fiber foods, such as whole grains, legumes, vegetables, and fruits, while the intake of refined grains (e.g., white rice and white flour) and high-sugar foods was reduced. Protein intake focused on high-quality sources, including fish, poultry, eggs, legumes, dairy products, and nuts, with limited consumption of processed meats. Fat sources were mainly unsaturated fatty acids, derived from olive oil, fish oil, nuts, and avocados, while intake of saturated and trans fats was minimized. This approach not only ensured appropriate macronutrient distribution but also optimized dietary quality, thereby improving insulin sensitivity and metabolic outcomes.

In this study, a 3-day food record was used to document patients’ dietary intake. Participants were instructed to consecutively record all foods and beverages consumed over 3 days (including two weekdays and one weekend day) within each 2-week follow-up period, detailing food types, portion sizes, and cooking methods. Dietary compliance was assessed based on the degree of deviation between actual dietary intake and the prescribed nutritional targets. During the 3-month intervention, patients were managed through a combination of regular follow-ups and daily monitoring. Patients attended clinic visits every 2 weeks, during which dietitians evaluated dietary compliance and physical indicators and modified dietary plans as necessary. In addition, these data were submitted to the research team through an online platform or telephone to support process management and compliance monitoring.

Baseline information of different BMI patients was collected, including age, duration of symptoms (years), family history of PCOS, cardiovascular disease and diabetes, and regularity of menstrual cycle (regular, irregular, and no menstruation) (17). The duration of a menstrual cycle may be classified as normal (21–35 days), brief (less than 21 days), or longer (exceeding 35 days). Menstrual flow can be divided into normal (30–80 mL, change sanitary pads or tampons every 4–6 h), excessive (more than 80 mL, change sanitary pads or tampons every 2 h or less, and last for 7 days), or insufficient (less than 30 m, change sanitary pads or tampons every 6 h or more) (18). The degree of ovarian enlargement was classified as normal (2–4 cm), mild enlargement (4–5 cm), and severe enlargement (>5 cm). The number of follicles was classified as normal (8–11), mildly increased (12–20), and significantly increased (>20) (19), while ovarian cysts are classified as no cysts (0–1), mild cysts (2–3), and severe cysts (>3). Clinical outcomes included ovulation rate, insulin resistance index (HOMA-IR), low-density lipoprotein (LDL) levels, high-density lipoprotein (HDL) levels, fast plasma glucose (FPG) levels, and BMI changes before and after nutritional intervention. After 6 months of treatment, the prognosis was defined as follows: ovulation rate reaching more than 65% (with ovulation occurring for 4 months or more), menstrual cycle being normal or close to normal, BMI decreasing by more than 5% (20), and fasting blood glucose levels decreasing by more than 10% (21). In this study, such outcomes were defined as an improved prognosis; otherwise, prognosis was considered poor.

Continuous data were represented as median (minimum–maximum) and categorical data were expressed as frequency (percentage). A multivariate regression analysis was used to evaluate the impact of factors such as BMI, age, duration of symptoms, family history of PCOS, and degree of ovarian enlargement on the effectiveness of nutritional interventions (with improved prognosis coded as 1 and poor prognosis coded as 0). Interaction terms and ROC curve analysis in the regression model were used to assess the impact of the interaction between BMI and other variables on the intervention outcomes. All statistical analyses were conducted using R software, and a p-value of <0.05 was considered statistically significant.

The age of PCOS patients ranged from 18 to 42 years, with a median age of 29 years, and the median duration of symptoms was 5 years. Regarding menstrual characteristics, the majority of patients presented with an irregular menstruation pattern or prolonged cycles, whereas regular or shortened cycles were relatively less common. More than half of the patients experienced a heavy menstrual flow. Ultrasound findings revealed that the majority of patients had mild ovarian enlargement with 12–20 follicles, although a considerable proportion exhibited severe ovarian enlargement or more than 20 follicles. In addition, more than 90% of patients had multiple ovarian cysts (Table 1). Among the 52 patients who dropped out, 25 were in the normal weight group, 15 in the overweight group, and 12 in the obese group, with comparable proportions across BMI categories. The age distribution was similar among the groups, as was the duration of symptoms. In addition, no significant differences were observed among the groups with respect to menstrual cycle patterns, including regular, irregular, and absent menstruation cycles (Supplementary Table 1).

Before treatment, there were no significant differences in the ovulation rate, HOMA-IR, LDL, HDL, and FPG among patients with different BMI groups. After treatment, the ovulation rate significantly improved in all groups (p = 0.038), with the obese group showing the most significant improvement. All groups showed a decrease in HOMA-IR (p = 0.040), indicating that treatment improved insulin resistance. During the treatment process, the obese group showed the largest change in BMI (−5.8%), and there were significant differences in this change among different BMI groups (p = 0.018). After treatment, LDL decreased in all groups (p = 0.043), HDL levels increased in all groups (p = 0.034), and fasting blood glucose decreased in all groups (p = 0.020) (Table 2).

The results showed that older age was associated with a slight decrease in intervention effectiveness (odds ratio (OR) = 0.993, p = 0.018). The longer the duration of symptoms, the worse the intervention effect (OR = 0.982, p = 0.039). Patients with a family history of PCOS showed significantly poorer intervention outcomes (OR = 0.745, p < 0.001). Patients with significant ovarian enlargement had poorer intervention outcomes (OR = 0.887, p = 0.003). The intervention effect was poor in patients with more ovarian cysts (OR = 0.882, p < 0.001). A higher BMI was associated with better intervention outcomes (OR = 1.089, p = 0.001). High HDL levels were associated with better intervention outcomes (OR = 1.010, p = 0.006) (Table 3).

The interaction with age indicates that the coefficient is negative (−0.008), which indicates that the relationship between BMI and age had an inhibitory effect on the occurrence of PCOS prognosis improvement. An OR value of 0.992 indicates that, for each additional unit increase of BMI–age combination, the likelihood of PCOS prognosis improvement decreased by approximately 0.8%, indicating thatthe impact of BMI on PCOS prognosis improvement gradually weakened with increasing age. For younger patients, nutritional intervention may exert different effects in patients with different BMI; however, for older patients, the impact of BMI on nutritional intervention becomes less significant. Similarly, the interaction term between BMI and ovarian size was statistically significant (p = 0.001), indicating that the combined effect of BMI and ovarian size had an inhibitory effect on the prognosis improvement in PCOS. This suggests that the degree of ovarian enlargement weakened the impact of BMI on nutritional intervention outcomes. In patients with less ovarian enlargement, BMI had a more significant impact on nutritional intervention outcomes, while in patients with more ovarian enlargement, the role of BMI was influenced by other factors such as ovarian dysfunction, thereby weakening its effect (Table 4).

Using ROC curves to further validate the results of the interaction analysis, the area under the curve (AUC) value for BMI was 0.728 in younger patients, while the AUC value was 0.643 in older patients (Figures 1A,B), indicating that BMI had a higher predictive ability for the effectiveness of nutritional interventions in younger patients. The AUC value of BMI was 0.697 in patients with less ovarian enlargement and 0.686 in patients with more ovarian enlargement (Figures 1C,D), indicating that nutritional intervention had significantly different effects on patients with different BMI groups in patients with less ovarian enlargement.