Introduction
Endometrial cancer (EC) is one of the most common gynaecologic malignancies worldwide, with an estimated 66,200 new cases and 13,030 deaths in the United States in 2023 [
1]. Despite efforts to combat the disease, the incidence of EC has increased by over 50% in the past 20 years [
2]. Approximately 70% of patients with EC are diagnosed at early stages due to the first symptom of uterine bleeding, and their 5-year overall survival (OS) might reach 80% [
3]. However, asymptomatic and advanced-stage patients have higher risk. Patients with metastatic disease, pelvic recurrence or extrapelvic recurrence further decrease the 5-year OS rates to 16%, 55% and 17%, respectively [
4,
5]. Accurate prediction and reproducible prognostication are therefore crucial for optimal treatment and outcomes.
Traditionally, tumour type, grade and stage are considered the most important prognostic factors in clinical evaluation. In recent years, many clinical tests and pathologic features following ESGO/ESTRO/ESP guidelines have provided clinicians with beneficial prognostic information and treatment recommendations [
6]. However, extensive testing increases the economic burden on patients and may not be suitable for long-term surveillance in undeveloped areas. Clinicians often overlook the potential of low-cost and reproducible biomarkers obtained from routine laboratory tests for cancer prognosis. In our previous study, we found that an elevated preoperative blood neutrophil-lymphocyte ratio (NLR), platelet-lymphocyte ratio (PLR), monocyte-lymphocyte ratio (MLR) and fibrinogen levels were significantly associated with poor OS in EC [
7,
8]. In subsequent studies, we examined standard tests used in clinical practice, including routine blood analysis, blood lipids, coagulation profiles, liver function indices, renal function indices and ABO blood type, aiming to incorporate these prognostic factors into an accurate and timely prediction model to improve patient outcomes.
Nomograms, which are being increasingly used as prognostic models in oncology [
9,
10], can graphically integrate diverse prognostic factors and provide a simplified and personalized scoring system for each patient [
11]. In this study, we first developed nomograms using a series of clinicopathological factors to predict risk of PFS and OS in untreated EC patients. Then we performed external validation to verify that the nomogram models integrating routine laboratory biomarkers are simple, reliable and effective in predicting outcomes for EC patients.
Discussion
Preoperative risk stratification is one of the greatest challenges in EC treatment [
14]. Conventional FIGO staging systems have limitations for predicting the prognosis of EC after surgery [
15]. Routine laboratory tests are simple and convenient for doctors to evaluate prognosis and are minimally invasive and cost-effective for patients [
16]. Therefore, establishing an objective predictive model incorporating laboratory indicators would be meaningful for evaluating the prognosis of EC. Nomogram model have reasonable and personalized prognostic value in facilitating management-related decisions [
17], integrating multiple clinical and biological risk factors into the evaluation and providing a visual, objective, and individualized scoring system for each patient [
11]. To verify our model’s predictive power, we established an independent validation set of patients. This method ensures the model’s generalizability for wide use in different patient populations [
18].
The nomogram developed in this study demonstrated excellent performance in predicting PFS and OS for patients with EC who had undergone comprehensive primary surgical treatment with total hysterectomy. Indeed, calibration curves displayed good discriminatory power in predicting 3-year and 5-year PFS and OS. The C-index and AUC in the training and validation sets indicated that the nomogram incorporating NLR, MLR, fibrinogen, albumin, and ABO blood type into the scoring system was reliable at predicting EC prognosis. The DCA for Model 1 also showed incredible net benefits, indicating that the nomogram involving the above indicators has good predictive power and clinical utility. Additionally, all calibration curves showed little deviation from the reference line, indicating high credibility. Thus, routine laboratory indicators are necessary for evaluating the outcomes of patients with EC. Furthermore, we categorized patients into three risk groups based on the prognostic nomogram scores using X-tile analysis. The findings revealed the risks associated with different scores and can help clinicians to intuitively predict the probability of survival.
This study examined a series of indicators collected from standard blood tests. Eventually, we found that high NLR, high MLR, high RDW, high fibrinogen, low albumin, and type AB blood were significantly associated with poor OS and PFS in EC. Our previous study described that tumour progression and systemic inflammatory responses disrupt the balance among routine blood constituents [
7]. Tumour cells induce an increase in neutrophils and monocytes, and in turn, neutrophils and monocytes inhibit the antitumour immune response [
19]. In contrast, tumour-infiltrating lymphocytes exhibit potent antitumour functions [
20]. Thus, increased NLR and MLR reflect the host’s immune status and might be associated with tumour progression and poor outcomes, as supported by similar findings in other malignancies [
21]. Fibrinogen and albumin are two acute-phase proteins induced in response to systemic inflammation but show opposite abundance trends under cancer inflammatory stimulation [
22]. Tumour cells directly synthesize fibrinogen and produce interleukin-6 to stimulate fibrinogen secretion, leading to tumour progression and metastasis [
23]. Albumin is an essential protein for nutrition transport and body metabolism, and hypoalbuminaemia is associated with poor outcomes in various tumours [
16]. Our previous study revealed that high fibrinogen and low albumin are significant prognostic factors for EC patients [
8]. The ABO blood group is also associated with EC prognosis. One hypothesis suggests that under normal conditions, von Willebrand factor (vWF) stabilizes factor VIII (FVIII) and transports it to injury sites, interacting with platelets and promoting the clotting process. The function of vWF is partly regulated by metalloprotease, which clears vWF from the plasma. The A and B antigens interfere with cleavage sites, reducing clearance of vWF. People with the AB blood group have the highest levels of vWF and FVIII in their plasma, which puts them at the highest risk of venous and arterial thromboembolism [
24]. Additionally, existing evidence suggests that cancers are often associated with a hypercoagulability state [
25]. These findings are consistent with many published studies showing a close association between the prognosis of various cancers and patient nutrition and immune status [
26]. The underlying mechanism may be that chronic systemic inflammation in patients depletes available nutrition and energy, leading to hypoalbuminaemia or even cachexia, resulting in poor outcomes [
27].
In some cases, PLR, RDW, and TG/HDL-c have been recommended as prognostic markers for EC [
28‐
30]. However, after considering dozens of routine laboratory biomarkers, they were found to be significantly associated with OS and PFS in univariate analysis but not in multivariate analysis. The prognostic values of those markers may not be sufficient when considered together, which is why they were not significant in multivariate analysis. Nevertheless, we cannot deny their association with EC prognosis, and it is essential to be cautious if their values are abnormal. Overall, the significant biomarkers identified in multivariate analysis deserve more attention to establish a precise and reliable prognostic evaluation system.
To the best of our knowledge, this nomogram model integrates the most comprehensive laboratory biomarkers in China. The findings are in line with existing studies and the model performs good robustness via external validation. However, there are still some weakness. First, as this study focuses on routine laboratory biomarkers, some prognostic markers were not included, such as tumour sizes and blood group antigens [
31,
32]. In addition, histopathological evaluation has a strong association with EC prognosis and is generally taken as a cornerstone for EC classification. The ESTRO/ESGO/ESP proposes four molecular TCGA molecular groups, namely, POLE ultramutated (POLEmut), mismatch repair-deficient (MMRd), p53 mutant (p53abn), and others referred to as NSMP (non‐specific molecular profile), to assess the prognosis of EC [
6,
33]. In some cases, molecular subgroups have been integrated into prognostic models to evaluate the association of other clinicopathologic factors with EC [
34]. It’s a pity that molecular marker detection was not extensively used in our institution during data collection. The nomogram is an inclusive model, and we will integrate more clinicopathologic factors to make the model more precise. Secondly, although we made external validation to mimic new patient cohorts, it is still a single-institution study. We will unite more institutions to improve the application universality and prediction accuracy.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.