The major results of this study showed that the nomograms demonstrated good performance and accuracy in both training and validation cohorts, and their prediction was supported by C-index, calibration, ROC curves, and DCA. When compared with the AJCC staging systems, the nomograms showed a better accuracy for survival prediction.
Age, CRC radiation, grade, histology, SEER stage, treatment, and tumor size consisted of the OS prediction model. In comparison with most other nomograms of FPEC, our findings firstly supported CRC-based radiation was negatively correlated with the OS of SPEC patients. Especially, evaluations of the underlying risk of second primary cancer (SPC) after radiotherapy for pelvic cancers, including prostate[15], bladder[16], cervix uteri[17], and rectum[3, 4], have substantiated the idea of an increased incidence of SPEC in patients undergoing pelvic radiotherapy, regardless of 1-, 5- or 10-year incubation period[3, 18]. Additionally, Guan et al indicated that 10-year survival rates in patients with radiotherapy-associated SPC are less than that in matched patients with first primary cancer[3]. Collectively, the variable of CRC radiation played a significant role in the nomogram of OS. Other than CRC radiation, other significant predictors of our OS nomogram, one or more, were adopted for various previous studies in different populations of EC. This phenomenon seemed to be partially attributed to the similar clinicopathological features between FPEC and SPEC in Japanese patients reported by Keiichiro Nakamura et al, which also implied the reliability of clinical interpretation of the OS nomogram from another aspect[19]. Unlike the large sample model with emphasizing a major impact of age at diagnosis on patient survival constructed by Sun et al[20], our study did not adjust for age at diagnosis. To minimize the offset due to the impact of the survival rate of patients with EC with time over and medical technology advances, we reclassified the tumor stage of all cases according to the 8th edition of the AJCC staging system, added the SEER summary stage, and modified pathological grading based on previous reports, because the period of case selection spanned two SEER databases from 1973 to 2020.
A little controversial finding was that age did not remain an independent predictor of CSS in this study, in contrast to many studies on EC where known age is a crucial prognostic factor. However, the present study participants were predominantly over 60 years old, above the average age from 58–61 years of FPEC[21], compared with these similar age distribution-based studies. Consistent with the work by Fleming et al., which showed that age greater than 70 in patients with EC was not a statistically significant predictor of poor outcomes after adjusting for other poor prognostic variables[22]. Previous research on patients with low CLDN6 expression also supports our finding[23]. On the other hand, the CSS model did not incorporate CRC-based radiation. One likely explanation is that patients who died from EC had a more aggressive histological type and more advanced disease, thereby masking the true underlying impact.
A significant similarity of both nomograms was the equal importance of treatment features supporting survival benefits, although treatment data solely relied on broad categories. Additionally, the survival outcomes of surgery combined with radiotherapy and chemotherapy appeared to be greatest by unadjusted survival analysis. Nevertheless, further risk stratification tests suggested not all patients can benefit from multimodality therapy. we discovered that patients who received radiotherapy plus chemotherapy had higher survival rates than other treatments in the high-risk group, but for the low-risk group, surgery plus chemotherapy was the better choice. This finding might at first look appear to be incongruent with the traditional therapeutic methods mainly by surgery, supplemented by radiotherapy and chemotherapy. However, the strategies of treatment for EC have been alterable due to the heterogeneity of high-risk populations. Gynaecologic Oncology Group (GOG) 122 is a pivotal study that changed the way we think about EC and chemotherapy[24]. A randomized trial included women with early-stage high-risk and stage III EC and reported significant improvements in recurrence-free survival and OS trends from chemotherapy plus radiation compared with chemotherapy alone [25]. On the contrary, Professor Daniela Matei’s team showed that chemotherapy plus radiation did not significantly improve relapse-free survival rate in terminal EC, but distant metastasis controlled[26]. Briefly, adjuvant radiotherapy might have an impact on those patients who had chemotherapy as primary treatment, which seemed likely to explain our findings that high-risk EC patients benefited most from chemoradiotherapy. Moreover, radiotherapy has been the standard adjuvant treatment in endometrial cancer historically. But lately, chemotherapy has emerged as a primary treatment alternative in early endometrial cancer[27]. Adjuvant radiation use is greatly reduced on account of the results of postoperative radiotherapy in endometrial carcinoma (ORTEC)-1[28]. The GOG-122 study showed that postoperative adjuvant chemotherapy was superior to whole adjuvant abdominal irradiation for EC patients, consistent with multiple randomized controlled trials[29–31]. Even, adjuvant chemotherapy could delay metastases, and the side effects could be acceptable, although chemotherapy failed to prevent local recurrence, also pointed out by these trials. Notably, our results could partially be attributed to the population differences between the target patient with SPEC in CRC and the FPEC although randomized studies on comparison of the treatments of patients with SPEC are worth further exploration. More individualized treatment was therefore suggested for EC patients after CRC.
There are several limitations of this study. First, since our nomogram was based on the data of Western countries, it may not be generalizable to all countries. Second, the prognostic model had not been externally validated, and practical advantages for prospective clinical use remain unclear. Third, this current study ignored the effect of diagnosis year, which may have an unexpected impact on patient outcomes with the advancement in medical technology over time. Finally, the relatively small sample size may limit statistical power, further research with large sample sizes is necessary to verify.
In summary, in this study, two objective and accurate prediction nomograms as drawn up informed patients about their prognosis and provided oncologists some reference for clinical decision-making based on the subpopulation at different risks. More works are required to confirm whether it has potential generalization ability in other patient groups.