Nomogram Predicting Overall Survival After Surgical Resection For Retroperitoneal Leiomyosarcoma Patients

Aim: To develop a survival nomogram for patients with retroperitoneal leiomyosarcoma (RLMS) after surgery. Methods: 118 patients with RLMS after surgical resection at the General Surgery Department, Shanghai Public Health Clinical Center, Fudan University were retrospectively analyzed. The nomogram was constructed based on COX regression model and discrimination was assessed using the concordance index (c-index). The predicted and actual survival was evaluated through calibration plots. Results: The c-index of the nomogram was 0.779 (95% CI, 0.659-0.898). The predicted and actual survival probabilities are in good agreement in all calibration curve. Conclusion: This study built the rst survival nomogram for patients with surgical resected RLMS. This study offers a RLMS-specic postoperative nomogram to assess for RLMS after It improves current RLMS staging with divided patients into low-risk and high-risk group. With the advantage of focusing on the RLMS, this tool can assist in


Introduction
Leiomyosarcoma (LMS) is a mesenchymal malignant tumor with distinct smoothmuscle differentiation [1]. accounting for about 7-10% of all soft tissue sarcoma (STS) and is the one of the most common STS subtype [2]. Currently, surgery is the best treatment for LMS that provides cure. But about 70% patients suffer from disease recurrence within ve years after surgery [3].
Nomograms are a statistical tool used to predict an individual's prognosis and have been widely applied in LMS [4][5][6]. But French Sarcoma Group's reported that, LMS represent a heterogeneous group of tumors, and RLMS having totally different clinical outcome and molecular features than others [7]. Though it has been reported that the tumor size, resection margin status and tumor grade are related risk factors for the prognosis of retroperitoneal leiomyosarcoma [8][9][10]. However, as far as we know, there is currently no nomogram which constructed to predict the survival of RLMS patients only.
Therefore the aims of this study were (1) to investigate the independent prognostic factors for the overall survival (OS) of RLMS patients and after surgical resection (2) to construct a nomogram to predict the 1year, 2 years and 5years OS of individual RLMS patients.

Patients
The study was approved by the Ethics Committee of Shanghai Public Health Clinical Center. All enrolled patients signed an informed consent form for data collection during their hospitalization. 118 consecutive patients who undergone abdominal surgery only between September 2010 to December 2020 at the Sarcoma Center, General Surgery Department, Shanghai Public Health Clinical Center, Fudan University, Shanghai, China were included. The inclusion criteria were as follows: (a) complete clinical information, (b) complete follow-up information, (c) histologically con rmed leiomyosarcoma, (d) the primary site located in the retroperitoneum. Leiomyosarcoma is de ned by the characteristic pathological features of H&E staining, including oval or cigar-shaped nuclei with blunt ends, variable eosinophilic cytoplasm, and uniform α-sma, desmin, and/or h-caldesmon positivity staining, and all GISTs (CD117+ and CD34+) were excluded [11].
Patient and tumor variables, included gender, age (<60 or ≥60), metastatic disease (patients with concurrent distant metastases or not), times of surgery (initial, second or more), complete resection (complete: R0, negative microscopic margin; incomplete : R1, positive microscopic margin or R2, macroscopically incomplete), number of resected organ (0-1 or >1), tumor burden (the sum of the largest diameters of all tumors), Federation Nationale des Centres de Lutte Contre le Cancer (FNCLCC) grade [12], multifocality, Inferior vena cava invasion, radiation (including preoperative and postoperative), chemotherapy (including preoperative and postoperative), other therapy (intervention, radiofrequency, gamma knife and intraperitoneal hyperthermic perfusion chemotherapy). The treatment plan for all patients is determined by a multidisciplinary team, including specialists in surgical oncology, medical oncology, radiation oncology, and pathology. Adjuvant treatments were used on a case-by-case basis. Each postoperative follow-up requires clinical and CT or MRI imaging of the chest, abdomen and pelvis.
The initial follow-up is every 3-4 months, then every 6 months after 2 years, and every year after 5 years.

Statistical analysis
Statistical analyses were performed with SPSS version 22.0 (Chicago, IL, USA) and R software (version 3.1.1; http://www.r-project.org/) with the "rms", "foreign", and "survival" packages. The primary end-point of the study was OS (de ned as the time from date of the surgery to the time of death or last contact). OS was estimated by the Kaplan-Meier method. The independent risk factors for OS were determined by Cox regression model. Variables with p-values <0.2 in univariate analysis were included in multivariate analysis. Then variables in the multivariate analysis with a p value < 0.1 were used to construct the nomogram to predict 1-,2-and 5-year OS after surgery. The predictive accuracy of the nomogram was assessed by calculating the c-index and drawing a calibration curve to judge the discriminative ability.
Then internally validated by Bootstrap method. Finally, the cohort was divided into low-risk group and high-risk group based on the median score calculated by the nomogram. The OS of two group was estimated by the Kaplan-Meier method, and the difference was compared with the log-rank test. All the tests were two-tailed and P < 0.05 considered statistically signi cance.
In the univariable analysis, variables including number of resected organ (p=0.012), tumor burden (p=0.005) and FNCLCC grade (p<0.001) were related to OS (

Nomogram development and validation
Variables with p<0.1 in multivariate analysis (number of resected organ, tumor burden, FNCLCC grade and multifocal disease) were used to construct the OS nomogram for patients with RLMS who had the abdominal lesions removed (Figure 2).
The c-index after internal validation was 0.779 (95% CI, 0.659-0.898). The calibration curves of 1-year, 2year, and 5-year OS probabilities show the predicted and actual survival probabilities are in good agreement ( Figure 3).
The nomogram gives a speci c value to quantify each variable. According to the total score, 118 patients were divided into low-risk group (<100) and high-risk group (≥100). The median OS was 116.5 (95% CI, -) and 30.0 (95% CI, 24.2-35.8 months) months, respectively. There was a statistically signi cant difference between the two groups (p<0.001) (Figure 4).

Discussion
Since the statistical prediction model can be simpli ed into a continuous numerical estimate tailored to individual patient conditions, the role of the nomogram prediction model has gradually become prominent in the era of precision medicine. In order to improve the estimation of the prognosis for patients with subtype pathology, some nomogram prediction models have been incorporated into the AJCC staging system. In the eighth edition of the AJCC manual, the nomogram by Trans-Atlantic Retroperitoneal Sarcoma Working Group for retroperitoneal soft tissue sarcoma (RPS) [13] was included as a model that met all AJCC quality criteria.
There have been many reports on nomogram prediction models for leiomyosarcoma. For example, MingFeng Xue et al. used the SEER database to establish a nomogram prediction model based on 1528 patients with extremity leiomyosarcoma, which can predict 5-and 10-year OS and cancer speci c survival, and The C-index values for internal validation of OS and cancer speci c survival prediction were 0.776 (95% CI 0.752-0.801) and 0.835 (95% CI 0.810-0.860), respectively [4]. Oliver Zivanovic et al. used a cohort of 185 cases of uterine leiomyosarcoma to establish a nomogram prediction model that can predict 5-year OS, with a c-index of 0.67 (95% con dence interval, 0.63-0.72) [6]. However, not only speci c STS histology but also the site of origin maters in determining outcomes in patients with soft tissue sarcoma. But there is currently no nomogram prediction model focused on retroperitoneal leiomyosarcoma only. Therefor, we developed and internally validated a novel, RLMS-speci c nomogram for predicting 1-, 2-and 5-year OS, and the c-index of our nomogram was 0.779 (95% CI, 0.659-0.898), the calibration plots shows that the predicted OS rate was perfectly match with the actual OS rate.
RLMS is a very rare mesenchymal malignant that originates in the retroperitoneal space. Its incidence rate is less than 1 per million population [2]. For this usually fatal but rare disease, the question of how to obtain long-term survival is very important. Because of metastasis, multifocal disease, and multiple organ resection are usually contraindications to surgery, the potential bene cial effects of surgery are not always obvious, which lead decision-making very complicated. Because of metastasis, multifocal tumor, and multiple organ resection are usually contraindications to surgery [14]. And it is extremely di cult for clinical decision-making whether or not patients with recurrence should undergo surgery. However, we can get some hints in the nomogram established in this research. For example, for those diseases where the preoperative assessment does not require combined organ resection, tumor burden is less than 5 cm, FNCLCC grade 1 and single center disease, even if it is a multiple recurrence disease, the 5-year OS rate after may exceed 90%, and the operation is obvious pro table. On the contrary, those who need combined resection of multiple organs, tumor burden is greater than 5cm, FNCLCC grade 3, multifocal disease, the nomogram score is greater than 200 points, and the OS is less than 30% at 2 years after surgery. For such people, the choice of surgery needs to be more cautious.
In the survival analysis, we found that FNCLCC grade and multifocal disease are independent risk factors for postoperative OS. The FNCLCC system is a commonly used histological grading system for soft tissue sarcomas, and it is one of the best indicators for predicting metastasis-free survival and OS [15].
This study is similar to the previous study reported by Qian Li et al., for RLMS patients with recurrent or metastatic disease who had a higher FNCLCC grade experienced worse prognosis [16]. Consistent with previous reports on RPS, patients with multifocal disease accounted for 23% of patients in this study [17], and compared with single-center disease, the risk of death in patients with multifocal disease increased by two times. Although there have been reports about the role of multifocal disease in the prognosis of RPS, as far as we know, this research is the rst to report its a independent prognostic factors in RLMS.
This study had certain limitations. First, this study was based on a retrospective cohort. The selection and inclusion of retrospective patients may cause research bias. Second, the median follow-up time for surviving patients in this study was 31 months, and further extension of the follow-up time will increase the reliability of the data. Third, although internal veri cation shows that this nomogram was a good predictive model, it still needs to further incorporate data from other centers for external veri cation.

Conclusions
This study offers a RLMS-speci c postoperative nomogram to assess OS for RLMS patients after surgery. It improves current RLMS staging with divided patients into low-risk and high-risk group. With the advantage of focusing on the RLMS, this tool can assist doctors in clinical decision-making.  Figure 1 Overall survival in patients with retroperitoneal leiomyosarcoma Figure 2 Nomogram for 1-year, 2-year and 5-year overall survival in patients with retroperitoneal leiomyosarcoma  OS curves strati ed by the score calculated by the nomogram and was strati ed according to the risk score as follows: low-risk group (<100) and high-risk group (≥100)