GBC is a prevalent malignant tumor of the biliary tract with poor prognosis [1, 4]. Currently, early-stage GBC is commonly diagnosed by pathological examination after cholecystectomy [16–18], defining as “incidental GBC” [19], and the corresponding prognosis is relatively favorable compared to those at advanced stage. Due to the lack of apparent clinical manifestations and early screening methods at the early stage of GBC, most patients were diagnosed of advanced stages, including distant metastases [5, 6]. These patients’ overall prognosis were relatively poor and aggressive surgical or comprehensive treatments were not strongly recommended. However, whether different clinical treatments can benefit for the prognosis of patients with distant metastatic GBC, and how to screen the corresponding candidate patients remained to be further explored.
Individualized treatment is considered important for malignant tumors, since clinicians can provide optimal treatment modalities according to various factors including patients’ demography data, tumor morphology and staging [20]. Nomograms have been used for prognostic prediction in various tumors such as hepatocellular carcinoma, intrahepatic cholangiocarcinoma, gastric cancer, breast cancer, lung cancer, cervical cancer [21–27]. Quantifying the independent prognostic factors and therefore calculating individualized predicted survival rates were the most important advantages of nomogram. A series of nomograms for the prognosis predictions of GBC have been constructed, while no study existed for those with distant metastases, and the potential prognostic factors were not comprehensively evaluated. Bai et al established a nomogram to predict the OS of GBC patients based on jaundice, CA199, staging and surgical margin, while lymph nodes were not included for evaluation [28]. He et al established a nomogram for GBC prognosis prediction using the SEER database, while none of surgery or lymph node dissection was referred, though the N staging was included as a variable [29]. In another study, lymph node dissection was adopted as a variable, while other treatments such as chemotherapy, radiotherapy were not taken into consideration [30]. In the present study, patients with distant metastatic GBC were enrolled, and the potential prognostic factors including patient demography, tumor morphology, staging, multiple treatment modalities such as surgery and comprehensive therapies were analyzed, which provided comprehensive clinical evidence for the prognosis predictions of these patients based on treatment benefits.
Comprehensive treatments including surgical and non-surgical were important for advanced GBC patients in terms of OS and quality of life, since patients might benefit from the reduction of tumor burden and the amelioration of tumor related complications. However, the significance of comprehensive treatments had not been fully elucidated in previous studies [31–34]. In this study, survival analyses were conducted according to different treatment modalities, and superior survival benefits were seen in the treatment cohorts (Fig. 1–2), which indicated that these patients might benefit from positive treatment modalities in candidate patients. Two prognostic nomograms were established based on the independent prognostic factors including or not including treatment factors (Fig. 4), and the validation efficacy was superior when treatment factors were included (Fig. 6), which also indicated potential survival benefits for positive treatment modalities in candidate patients.
Internal and external validations are crucial to evaluate the predictive efficacy of the established nomograms. Absence of external validation was common in similar studies due to the difficulty of acquiring elaborate clinical data. In the present study, clinical data were collected from the authors’ hospital according to the standards of SEER database, and were used for external validation. The calibration and discrimination capabilities were favorable in the external validation cohort, though the patient number was relatively small (Fig. 4–5). Specially, the discrimination capability of the established nomogram was superior to the 8th AJCC TNM staging system (Fig. 5, Table 3), indicating a better clinical significance. DCA is a commonly used method to evaluate the clinical applicability of nomogram [35, 36]. However, absence of DCA was not rare in previous studies, which weakened the clinical application value of the established nomograms, though the calibration and discrimination capabilities were favorable [30, 37]. The present study showed good DCA results, especially for the nomogram B, which was established based on prognostic factors including treatment factors (Fig. 6). Finally, risk stratification analyses using the average score showed a significant survival difference, which further indicated the good clinical applicability of the established nomogram (Fig. 7).
There also existed some limitations in this study. The SEER database did not include detailed treatment information and patient-related laboratory indicators, further analyses were not available. Besides, the number of enrolled patients for external validation was relatively small from a single center, which weakened the validation efficacy, including the clinical applicability. Large population and multicenter based prognostic factor analyses and subsequent predictive nomogram constructions and validations were needed in future studies.