Diabetes mellitus indicate poor prognosis in uterine serous cancer patients

Purpose. We aimed to nd prognostic factors for uterine serous cancer(USC) patients in a retrospective study. Methods. 51 USC patients between 2010-2020 were enrolled. All pathological specimens were reviewd. The research protocol was approved by Institutional Review Board and all patients were informed consent before the study began. Statistics were done using SPSS 25.0, T test and chi-square analyses were used to compare differences, the overall survival(OS) was estimated with Kaplan-Meier(KM) analysis, univariate and multivariate Cox regression analyses were utilized to nd prognostic factors. Results. The median overall survival(OS) and progressive free survival(PFS) were 75.94 and 63.49 months, respectively. Diagnosed with diabetes mellitus(P=0.006, HR=6.792, 95%CI=1.726-26.722) and CA125(cid:0)28U/ml(P=0.006, HR=7.136, 95%CI=1.780-28.607) before surgery were independent risk factors for OS, advanced FIGO stage(P=0.001, HR=10.628, 95%CI=2.894-39.026) and DM(P=0.003, HR=6.327, 95%CI=1.875-21.354) were independent factors for PFS. Age ≤ 52, , tumor size ≥ 2.5cm and cervical mucosal inltration may indicate poor prognosis but were not independent risk factors. Hypertension patients with routine medical treatment tend to survive longer, but there was no statistical differences in OS and PFS compared to patients with normal blood pressure. Conclusion. In addition to surgical and adjuvant treatments, gynecologists should focus more on the comorbid conditions of USC patiens, especially for DM.


Introduction
Endometrial cancer(EC) is the 7 th most common cancer and the 14 th most common cause of cancer death in women, globally [1]. Most EC presented as early-stage disease, with a favorable 5-year overall survival(OS) of greater than 85% [1]. However, uterine serous carcinoma(USC), which is classi ed into type 2 EC, is an aggressive and rare subtupe of EC with an estimated 5 year OS of 18-27% [2]. USC patients usually suffered from recurrence even in early stage. According to the FIGO guideline, the standard system treatment strategies are surgical staging(for patients with early stage) or debulking surgery(for advanced stage) followed by radiochemotherapy [3]. Improved systemic treatment protocols have prolonged the disease-free survival but have had little effect on long-term overall survival. We aim to nd risk factors of disease prognosis in USC patients, which would bene t survivors in providing consulting and management after primary treatment.
Co-morbidities may affect patients' pathological state and prognosis to some extent. Roy showed, cancer survivors had a higher prevalence of hypertension, diabetes mellitus(DM) compared to health people,and among all the cancer type, uterine cancer had the highest prevalence of DM(42.9%) [4]. Recent evidence suggested, DM may perform a more important biologic link in terms of uterine cancer [5,6]. Uterine cancer can be classi ed into type 1 and 2, the former was associated with high estrogen status and insulinresistance, including metabolic syndrome and DM [7]. In 2013, Zanders et al. showed a lower ve-year OS rate of 68% compared to 84% in patients with diabetes and EC [8]. Additional meta-analysis by Liao et al.
in 2014 showed a similar pooled relative risk of disease speci c mortality of 1.32 [6]. USC is type 2 uterine cancer which had a signi cantly poorer prognosis than type 1 uterine cancer, there were no studies on DM and USC.

Materials And Methods
A total of 51 patients diagnosed with USC between 2010-2020 were enrolled in this retrospective study, which was approved by hospital institutional review board. Inclusion criteria were as follows: (1) all patiens' pathological specimens were re-reviewd and recon rmed, (2)received primary operation and rstline adjuvant therapy.
Clinical informations were obtained from medical records. Patients were restage acording to the 2018 FIGO stage. All patients were initial diagnosed and operated in the Memorial Hospital of Sun Yat-sen, Sun Yat-sen University, followed by adjuvant therapy. Information of reccurrence or death were obtained from the follow-up department. In this study, all death was caused by USC, so overall survival was equal to cancer-speci c survival(CSS).
Statistics were done using SPSS 25.0(IBM, Armonk, NY, USA). The best cut off value of CA125, tumor size and age were estimated with ROC curve. T test and chi-square analyses were used to compare clinical variables between groups. Signi cance level is 0.05. The OS and PFS was calculated using Kaplan-Meier(KM) analysis and compared using the log-rank test. Univariate and multivariate Cox regression analyses were utilized to nd prognostic factors.
The Kaplan-Meier(KM) curves and forest plots were visualized via R 3.6.1, 'survival' and 'survminer' packages were used.

Results
Of the enrolled 51 patients, 16 had DM. Cut-off values for age, menopause time, CA125, and number of resected lymph nodes were calculated via SPSS using ROC curve.
Most patients(42/51) were dignosed with USC over 52 years old. About 2/3 patients were at early stage.
16 patients didn't receive lymphadenectomy, so we couldn't con rm their pathological status of lymph nodes.
The baselines of patients with or without DM are listed in Table 1. The differences between all the listed factors including stages, BMI, age, and etc were not statistically signi cant in the 2 group of patients.
The median PFS and OS of all patients were 63.49 months and 75.95 months, respectively( Table 1). The 2-year-OS rate was 85.4% and 2-year-PFS rate was 77.3%. The OS and PFS rate of patients with DM were signi cantly lower than those without DM(P=0.002 and 0.029, respectively, Fig 1a and Table 1).
To analyze pathological parameters and co-morbidities in USC patients, univariate and multivariate cox regressions were run. As shown in Table 2, DM was related to a poorer OS. DM and higher stage, BMI, CA125 were unfavourate factors for USC patients' PFS, small tumor size, higher number of resected lymph nodes, and negative lymph nodes were favourate factors for PFS.
In order to nd more potential risk factors for OS in multivariate regression analysis, we also include factors with P value≤0.10 and stage in multivariate analyses. We found that diagnosed with diabetes mellitus(P=0.006, HR=6.792, 95%CI=1.726-26.722) and CA125 28U/ml(P=0.006, HR=7.136, 95%CI=1.780-28.607) before surgery were independent risk factors for OS, the survival curve of these risk factors were shown in Fig  Interestingly, it indicated that hypertension may be a protect factor for USC patients, which contradicts with mainstream view, but the result was not statistically signi cant (Fig 1).

Discussion
USC is a rare but deadly subtype of EC, we conducted a single center retrospective study of 51 patients and found that DM was the independent risk factor of both OS and PFS, the HR were 6.792 and 6.327, respectively.
In our data, about 60% patients were early stage, which differed from previous study [9]. Almost all patients' complaints were vaginal bleeding or abnormal discharge, which may be the reason for the higher early stage rates of diagnosis.
Only 62.7%(32/51) patients had an accurate pathologic diagnosis before surgery, the rest of patients were mis-diagnosed with endometrioid adenocarcinoma by pre-operation biopsy. According to the famous 'Mayo Criterion', for patients of grade 1-2 stage I EC, with tumor size 2cm and myometrial invasion depth 1/2 regional lymphadanectomy could be avoided [10]. But for early stage USC, lymphadanectomy is debatable. In our study, Cox regression showed that although the number of lymph nodes removed was not an independent factor, patients with more lymph nodes resected had a better PFS. As for OS (Figure 2), K-M analysis showed patients without lymph node metastasis had a signi cantly better prognosis, and the OS of patients whose lymph nodes were not removed was poorer, patients with lymph nodes metastasis had the pooerest end. Li's study enrolled 2853 patients registrated in SEER, the result suggested that patients with ≥12 pelvic lymph nodes (PLNs) removed had signi cantly better OS than those without (both P < 0.001) [11]. Our results showed the number of lymph nodes removed was not associated with prognosis, we speculated that the small sample size was the reason.
Hence, we think systemic lymph node resection may be recommended to USC patients. Accordingly, the accuracy of pathologic diagnosis before operation was important, cause it directly impacted the decision to ressect lymph nodes or not. If we mis-diagnosed an early stage USC as an endometrioid adenocarcinoma, we may make a wrong surgical decision.
DM has reached epidemic preportions worldwidely, its rate shockingly doubled over the last 3 decades [12]. From an gynecologic cancer prospective, the rise in DM impacts patients' medical condition across their lifespan. A direct link between DM and type 1 EC is now well appreciated [13,14]. Hyperglycemia promoted the development of type 1 endometrial cancer and led to poor prognosis. High blood glucose facilitated EC tumor growth by providing carbon sources for diverse metabolic pathways [13]. In vitro experiments exhibited, the proliferation,adhesion and chemo-resistance of tumor cells were promoted when cultured with high glucose condition [15][16][17]. There are signi cant differences in molecular phenotypes and hormone receptor expression between type 1 and type 2 EC, thus the prognostic risk factors of the two types may be different. So did DM affect USC patients' survive time?
The answer was positive. Our study revealed for the rst time that DM was an independent risk factor for OS and PFS in USC patients. Shou did a retrospective analysis in 139 Chinese non-endometrioid adenocarcinoma patients(including 45 USC), he found that metabolic syndrome(MS), stage and age were independent risk factors for OS of USC patients [18]. MS was a group of diseases including DM, hypertension and hyperlipidemia, while our study was focus on DM only.
Surgery and radiochemotherapy are important in USC, but before or after these oncology treatments, we should pay more attention to the control of co-morbidities to improve outcomes. Metformin is commonly used to treat type 2 DM, it may bene t outcomes of EC in an indirect manner [19]. Perez found among postmenopausal women with EC, taking metformin could decrease overall mortality[20]. Sarfstein's study showed that metformin decreased expressions of IGF pathway molecures via inhibition of IGF-IR promotor activities in USC cells, nally promoted cell apoptosis, attanuated cell migration and proliferation [21]. On the contrary, Lilia reported that merformin shortened lifespan of C.elegans and human primary cells when provided in late life [22]. But this research studied individuals with healthy metabolic status. As we know, the peak age of USC is later than type 1 EC, therefore, the use of metformin may require more careful selection. Future prospective trials are needed to explore its impact on USC.
Our present study also found that, contrary to previous ndings, the prognosis of hypertensive patients seemed to be slightly better than that with normal blood pressure, although the difference was not statistically signi cant. By increasing the sample size, the differences may be weakened or enlarged.
Limitations of this study are as follows: 1. Single-center retrospective study with low level of evidence, 2. The sample size is small, but USC is a rare disease, and Gynecologic Cancer InterGroup(GCIG) appeals that it should be separately managed and studied[23], 3. For some patients, the follow-up time is less than 2 years, which may lead to bias.
The advantages of this study is that the operation quality is guaranteed in the famous gynecological tumor center in China. It is the rst time that medical comorbidities have been found to in uence the prognosis of USC, which will guide the treatment of these patients.    Figure 1 Kaplan-Meier survival curves of USC patients a) The KM curve for patients with(blue curve) of without DM(yellow curve) (P=0.002). b) The KM curve for USC patients with CA125≤28U/L(yellow curve) and CA125 28U/L(blue curve) (P=0.0016). c) The KM curve for USC patients with hypertension(blue curve) was higher compared with patients without hypertension(yellow curve), but the difference did not reach statistical signi cance (P = 0.11). d) The KM curve of lymph node status for USC patients (P=0.01), the prognosis of patients without lymph nodes metastasis were better than the group of patients whose lymph nodes were not resected, patients with lymph nodes invasion suffered the worst prognosis