Are the indications for postoperative radiotherapy in the NCCN guidelines for patients with gastric adenocarcinoma too broad? A study based on the SEER database

Although the morbidity of gastric adenocarcinoma (GA) has declined in recent years, this disease remains the fourth most fatal malignancy in the world [1, 2]. At present, surgery is the only reliable cure for GA [3‐5]. Regional recurrence and distant metastasis are the main causes of reduced survival among patients with GA [6, 7]. However, postoperative adjuvant chemoradiotherapy can reduce the risks of regional recurrence and distant metastasis [8‐12]. Therefore, the rational use of postoperative chemoradiotherapy can improve patient prognoses. The effects of postoperative adjuvant chemotherapy have been widely recognized. In contrast, although the NCCN guidelines for GA that exhibits mitosis [13] clearly suggest that a postoperative pathological stage of T3–4Nx or TxN+ indicates the feasibility of adjuvant chemoradiotherapy after surgery, which means that postoperative radiotherapy is based on postoperative chemotherapy, a portion of patients who undergo radiotherapy experience severe postoperative complications [14‐16], and the curative effects of radiotherapy are influenced by various pathological factors. As a result, the indications for postoperative radiotherapy remain controversial [17]. Thus, the purpose of this study was to use the Surveillance, Epidemiology, and End Results (SEER) database to explore risk factors for prognosis for patients who undergo postoperative radiotherapy; perform stratified analyses of the effects of radiotherapy on postoperative disease-specific survival rate (DSS) for patients with GA; construct a nomogram to filter out patients who are unsuitable for postoperative radiotherapy; and provide a reference for the reasonable implementation of postoperative radiotherapy.

In recent years, the clinical efficacy of postoperative chemoradiotherapy involving postoperative radiotherapy after postoperative chemotherapy has been confirmed for certain gastric cancer patients in randomized controlled trials[14, 21, 22]. However, although indications for postoperative adjuvant chemotherapy have been widely recognized, indications for postoperative adjuvant radiation therapy remain unclear. It is generally believed that the latter indications are influenced not only by TNM stage but also by factors such as the number of lymph nodes dissected and pathological classification [23]. The current NCCN guidelines suggest that gastric cancer patients with pathological staging of T3–4 Nx or TxN + should accept postoperative radiotherapy; these guidelines use TNM stage as the sole determinant of whether postoperative radiotherapy should be performed and thus cannot be applied in clinical contexts. In the present study, the SEER database was used to investigate more accurate decision factors for postoperative radiotherapy and thereby determine stricter indications for this therapeutic modality. The results show that not all these patients require postoperative radiotherapy, and we suggest that high-risk patients should accept postoperative radiotherapy, whereas low-risk patients should not.

Defining risk factors that influence DSS is an approach for studying indications for postoperative radiotherapy. Thus, each of the indicators in this study was an independent risk factor for DSS according to Multiple Cox regression analysis. Prior studies have demonstrated that more advanced TN stage, larger tumor diameter and a tumor pathology classification of diffuse can increase the difficulty of R0 tumor resection and thereby affect patients’ postoperative DSS [24, 25]. Having fewer than 15 lymph nodes dissected can increase the possibility of postoperative lymph node recurrence [26, 27], which also decreases DSS. The results of this study are similar to those of previous studies. Furthermore, we established a nomogram for predicting DSS for patients who underwent postoperative chemoradiotherapy; after grading the two groups of patients, a cut-off for nomogram-predicted 5-year DSS was determined via recursive partitioning analysis. Subsequently, we stratified our analyses of DSS for patients in the two groups based on this cut-off (69 points). The results showed that for all patients, DSS was significantly better for Group R than for Group C. However, among low-risk patients, DSS did not significantly differ for the two groups, indicating that postoperative radiation did not appear to influence DSS for low-risk patients. This result led us to question whether patients in the low-risk group should be required to undergo postoperative radiotherapy. Previously, basic studies have reported that GAs are not sensitive to radiation; normal gastric mucosa cells will already be injured by the time that cancer cells have been exposed to a lethal dose of radiation. Moreover, viscera adjacent to the stomach, such as the liver, pancreas, and other organs, are sensitive to radiation injury. Given the invasiveness of surgery, which causes decreases in patients’ physical strength, additional radiation can often be difficult for patients to tolerate. Therefore, a significant number of patients experience serious complications after postoperative radiotherapy [13‐15]. The results of the present study indicate that for low-risk patients, DSS did not improve after postoperative radiotherapy. Therefore, we believe that postoperative radiotherapy is not appropriate for low-risk patients.

This study further verified the value of applying nomograms via decision curves. A decision curve is used as a simple mathematical model in which the loss function [28] is employed to examine the effectiveness of a statistical model for inferring the outcome of an event; such curves have been widely utilized to evaluate the usefulness and benefit of forecasting models [29‐32]. The results of this study showed that when the threshold probability is 0–58%, clinicians should make decisions regarding whether radiotherapy should be administered to patients for whom this treatment is suggested by NCCN guidelines. This approach can produce greater beneficial effects than having either all or none of these patients undergo radiotherapy. The threshold probability represents the clinician’s degree of confidence in postoperative radiotherapy. We believe that if a patient is unable to tolerate complications associated with postoperative radiotherapy, the clinician’s confidence in postoperative radiotherapy for patients with GA will be affected. Prior studies [13‐15] have indicated that approximately 48–54% of patients with GA are unable to tolerate complications that arise after postoperative radiotherapy; therefore, we believe that clinicians should have approximately 50% confidence that postoperative radiation will benefit patients. This belief is consistent with the scope of application of the decision curve.

This study involved a large sample and long-term follow-up, and the obtained results were verified and validated. Nonetheless, there are a few limitations. First, a retrospective study may inevitably incorporate bias. Second, although SEER data from a single centralized repository offered the opportunity to examine a large sample, the SEER database is missing certain data regarding characteristics that affect outcomes, such as the positive margin rate and postoperative complications. Consequently, the results of this study may be insufficiently accurate. Therefore, to confirm our findings, more rigorous results should be obtained from multi-center, prospective, large-sample clinical trials.