Prognostic nomogram for glioblastoma (GBM) patients presenting with distant extension: a seer-based study

Glioblastoma (GBM) with distant extension is rarely reported. We retrieved the data of GBM patients from the SEER database to identify the prognostic factors of GBM with distant extension and constructed a nomogram to predict the overall survival (OS) of these patients. The data of GBM patients between 2003 and 2018 were retrieved from the SEER Database. 181 GBM patients with distant extension were randomly divided into the training cohort (n = 129) and the validation cohort (n = 52) at a ratio of 7:3. The prognostic factors associated with the OS of the GBM patients were identified through univariate and multivariate cox analyses. A nomogram was constructed based on the training cohort to predict OS, and its clinical value was verified using the validation cohort data. Kaplan–Meier curves showed that the prognosis was significantly worse for GBM patients with distant extension than GBM patients without distant extension. Stage (GBM patients with distant extension) was independent prognostic factor of survival. Multivariate Cox analyses demonstrated that age, surgery, radiotherapy and chemotherapy were independent risk factors for OS of GBM patients presenting with distant extension. The C-indexes of the nomogram for predicting OS were 0.755 (95% CI 0.713–0.797) and 0.757 (95% CI 0.703–0.811) for the training and validation cohorts, respectively. The calibration curves of both cohorts showed good consistency. The area under the curve (AUC) for predicting 0.25-year, 0.5-year and 1-year OS in the training cohort were 0.793, 0.864 and 0.867, respectively, and that in the validation cohort were 0.845, 0.828 and 0.803, respectively. The decision curve analysis (DCA) curves showed that the model to predict the 0.25-year, 0.5-year and 1-year OS probabilities was good. Stage (GBM patients with distant extension) is independent prognostic factor for GBM patients. Age, surgery, radiotherapy and chemotherapy are independent prognostic factors for GBM patients presenting with distant extension, and the nomogram based on these factors can accurately predict the 0.25-year, 0.5-year and 1-year OS of these patients.


Introduction
GBM is the most common and aggressive type of glioma and is characterized by poor prognosis even after multimodal treatment involving surgery, radiotherapy and chemotherapy. The median survival time of GBM patients is only 15 months (Alifieris 2015), and the 5-year survival rate is a dismal 12% (McBain et al. 2021).
Primary GBM can progress to scalp metastases, pulmonary metastases, bone metastases, and vertebra metastases (Liu et al. 2020;Zhang et al. 2021;Li et al. 2020b). However, unlike common malignancies such as that of lung and breast cancer, distant metastasis of GBM is rare due to the low survival rates of the patients. Furthermore, unique 1 3 physical barriers in the brain, including the dura mater, the thickened basement membrane and the blood-brain barrier, also reduce the probability of distant metastasis of GBM (Ray et al. 2015;Mourad et al. 2005). In fact, only 0.2-0.4% of GBM cases present with systemic dissemination (Beauchesne 2011). Due to rapid relapse and continuous progression, patients with GBM do not survive long enough before neuraxis or systemic metastases can be detected (Lah et al. 2020). Lun et al. conducted a meta-analysis of 88 cases of extracranial GBM between 1928 and2009, and reported a poor prognosis. The median age of the patients was 38 years, and the median OS was 10.5 months. In addition, there has been a progressive increase of 0.7 months per decade between 1940 and 2000 in the interval from detection of extracranial metastasis and death (Lun et al. 2011). Another meta-analysis, including 150 glioblastoma/ gliosarcoma patients, showed the median OS from diagnosis of metastasis (OSM +) was 6 ± 0.8 months, and median OS from initial diagnosis (OSID) was 13 ± 2.4 months (Pietschmann et al. 2015). Furthermore, individual case reports, small case series and meta-analysis have all demonstrated the poor prognosis of GBM patients presenting with distant extension. Thus, it is critical to evaluate the clinical characteristics, prognostic factors and treatment strategies for these patients.
SEER is a database of high-quality information for cancer incidence and survival in the United States, and has been extensively used for clinical research related to cancer (Cronin et al. 2014) since it provides a platform to integrate minor case series and isolated institutional studies. In this study, we retrieved the data of GBM patients with distant extension from the SEER database, and identified the prognostic factors associated with overall survival. A nomogram was constructed based on these factors to predict the survival of GBM patients with distant extension, which can help clinicians to quantitatively estimate the clinical prognosis and make treatment decisions accordingly.

Patients and methods
We used SEER to collect 14,802 GBM cases of nervous system tumors diagnosed pathologically. After eliminating invalid and missing clinical information, 14,707 GBM patients were finally identified. Among the 14707cases, 181 cases with 'distant' tumor stage were labeled as 'GBM patients with distant extension', and the other cases were labeled as 'others'. The data of patients who were diagnosed with GBM between 2003 and 2018 were obtained from the SEER Database. The patients were included on the basis of the follow- , nasal cavity, nasopharynx, other direct extension outside CNS and posterior pharynx. Distant metastasis included carcinomatosis, distant metastasis with or without distant lymph node(s), and metastasis within CNS and CSF pathways (including "drop" metastasis, extra-neural metastasis, and metastasis outside the CNS). The data of age, race, sex, year of diagnosis, primary site, laterality, surgery, radiation, chemotherapy, tumor size, survival months, and vital status recode were included in the analysis. Given that the data on "CS tumor size" were not accessible after 2016, the data of "Tumor Size Summary (2016 +)" were used for patients diagnosed between 2016 and 2018 instead of "CS tumor size" data. According to the SEER Program Coding and Staging Manual 2021, the different codes represent the following treatment strategies: 00-none, no surgery of primary site and autopsy only, 20local excision of tumor/lesion or mass and excisional biopsy, 21-subtotal resection of tumor, lesion or mass in brain, 30-radical, total, gross resection of tumor, lesion or mass in brain, 40-partial resection of the lobe of brain when the surgery cannot be coded as 20-30, 55-gross total resection of the lobe of brain (lobectomy), and 90-surgery, NOS, laser interstitial thermal therapy (LITT). The primary site of the tumor was categorized as supratentorial (cerebrum, frontal lobe, temporal lobe, parietal lobe, occipital lobe), infratentorial (cerebellum, brain stem), and other (ventricle, overlapping lesion of the brain, NOS).
Kaplan-Meier curves were plotted to evaluate the prognosis of patients. The 181 GBM patients with distant extension were randomly divided into a training cohort (n = 129) and a validation cohort (n = 52) in a ratio of 7:3. Categorical variables were evaluated using the chi-square test or Fisher's exact test. The prognostic factors associated with the OS were identified by univariate and multivariate Cox analyses, and the HRs (hazard ratios) and 95% CIs (confidence intervals) were calculated by multivariate and univariate Cox proportional hazard models. A nomogram was constructed based on the significant prognostic factors using the RMS package in R version 4.1.0 (http:// www.r-proje ct. org/). Harrell's concordance index (C-index) was calculated to evaluate the performance of the nomogram. ROC (receiver operating characteristic) curves were plotted, and the area under the curve (AUC) was calculated to evaluate the predictive ability of the model. Calibration curves were plotted to assess the similarity between the predicted and observed survival. Finally, DCA curves were used to assess the clinical net benefit of the nomogram. All statistical analyses were performed using R software (version 4.1.0). All tests were two-sided, and p < 0.05 was considered statistically significant.

Survival analysis and Cox regression analysis of GBM patients
Kaplan-Meier curves showed that the prognosis was significantly worse for GBM patients with distant extension than others. (P < 0.05, Fig. 1). Univariate analysis and multivariate analysis showed that stage (GBM patients with distant extension) was independent prognostic factor of survival (P < 0.05 S1). The sorts of distant extension of 181 patients are shown in Table S2. Distant spread was determined by imaging, needle biopsy, operative report or diagnostic reports (S3).

Independent prognostic factors of OS
The results of the univariate and multivariable Cox analysis are summarized in Table 2. Multivariate Cox analyses demonstrated that age, surgery, radiotherapy and chemotherapy were independent risk factors for the OS. The Kaplan-Meier curves of the overall cohorts, and that of the age, surgery, radiotherapy and chemotherapy subgroups are shown in Figs. 1 and 2.

Construction and validation of the nomogram
The nomogram model was constructed on the basis of age, surgery, radiotherapy and chemotherapy to predict the 0.25year, 0.5-year and 1-year OS of GBM patients presenting with distant extension (Fig. 3). The probability of survival was calculated by adding the scores of the individual factors. The C-indexes of the nomogram for predicting OS were 0.755 (95% CI 0.713-0.797) and 0.757 (95% CI 0.703-0.811) for the training and validation cohorts respectively, which was indicative of the high predictive value of the nomogram. As shown in Fig. 4, the calibration curves for 0.25-year, 0.5-year and 1-year survival rates suggested good agreement between the predicted and actual observations. Similar consistency was observed in the validation cohort as well. The AUCs for the 0.25-year, 0.5-year and 1-year OS in the training cohort were 0.793, 0.864 and 0.867 respectively. In the validation cohort, the AUCs for 0.25-year, 0.5-year and 1-year OS were 0.845, 0.828 and 0.803 respectively (Fig. 5). DCA curves were used to illustrate the clinical effectiveness of the nomogram, and the results showed that the nomogram model has a good net benefit in predicting the 0.25-year, 0.5-year and 1-year survival of GBM patients with distant extension (Fig. 6). Taken together, the nomogram can accurately predict the survival of GBM patients with distant extension.

Discussion
GBM rarely disseminates beyond the primary tumor site or outside the central nervous system, although the incidence of distant extension has increased over the past decades. This can be attributed to greater awareness of the extension of staging, as well as significant advances in imaging   diagnostics such as computed tomography (CT), high-resolution magnetic resonance imaging (MRI) and PET-CT. Moreover, improved local tumor control and prolonged survival due to combined treatment (Kalokhe et al. 2012) may also contribute to distant extension (Di Carvalho et al. 2019). Our study showed that stage (GBM patients with distant extension) was independent prognostic factor of survival, and this subpopulation of GBM patients deserves more attention and research. Distant extension of GBM is mainly driven by CTCs (circulating tumor cells), CTSCs (circulating tumor stem cells), altered brain-blood barrier permeability, and mesenchymal transition (Georgescu and Olar 2020;Lombard et al. 2015). The prognostic factors and appropriate therapeutic strategy of GBM patients presenting with distant extension have still not been defined completely. Age is positively correlated with the incidence of GBM, and the incidence of GBM is highest in the 75-79 years age group, and declines in the 80-84 and 85 + years age groups ). Our results on GBM with distant extension were similar. Furthermore, age has been consistently identified as a decisive prognostic factor for most malignancies. Li et al. found that age portended a worse prognosis for GBM patients older than 65 years (Li et al. 2020a). Likewise, Wen et al. showed that elderly GBM patients had worse OS compared to younger patients, and the benefits of chemotherapy were greater in patients who were younger than 65 years or who underwent additional treatments (Wen et al. 2021). This may be due to a lower lesser response to radiotherapy and a marked tendency toward limited treatment options with advancing age (Kita et al. 2009). Our model also shows that patients presenting with distant extension who are of advanced age have a poor prognosis. However, some studies show that age is not a prognostic factor in patients with extracranial metastases (Pietschmann et al. 2015;Lun et al. 2011). Regardless, with an increasingly aging population, more data is needed to validate the impact of age on the prognosis of GBM, and further studies are warranted that focus on elderly GBM patients presenting with distant extension.
The standard treatment of GBM is surgery (maximal safe resection), followed by radiotherapy and chemotherapy (Weller et al. 2014). Recurrent GBM can also be safely treated using repeat LITT when surgery cannot be effectively performed (Muir et al. 2022;Thomas et al. 2016). Partialbrain fractionated radiotherapy with concurrent and adjuvant temozolomide after biopsy or resection is recommended after surgery for GBM patients younger than 70 years (Sulman et al. 2017), and the use of a more aggressive approach is associated with longer survival in elderly patients (Scoccianti et al. 2010;Zarnett et al. 2015). However, due to the scarcity of cases, it is difficult to confirm the standard treatment for patients diagnosed with GBM presenting with distant extension (Ray et al. 2015). Lun et al. observed that patients treated with surgery, radiation, chemotherapy, and cerebrospinal fluid shunting, if necessary, had higher mean survival when compared to patients undergoing less intense treatments (Lun et al. 2011). A meta-analysis also confirmed that surgical resection can reduce the risk of GBM metastasis, and chemotherapy and radiation therapy can prolong survival in patients with metastatic GBM (Cunha and Maldaun 2019). However, reports are scarce on a possible correlation between the extent of resection and prognosis in GBM patients with distant extension. Our multivariate analysis and nomogram also showed that both beam radiotherapy and chemotherapy can significantly improve the prognosis of GBM patients presenting with distant extension. Furthermore, our findings indicate that these patients would achieve a better prognosis if they accepted a greater extent of tumor resection. Consistent with this, a previous study reported that the extent of resection (biopsy, partial, and complete resection) in elderly (> 65 years) GBM patients correlated with progression-free survival (2.1/3.4/6.4 m, P = 0,000) and overall survival (2.2/7.0/13.9 m, P = 0,000) (Ewelt et al. 2011). Furthermore, patients who underwent gross total resection showed a 2-to 3-month improvement in overall survival (P < 0.001) relative to those who underwent subtotal resection (Noorbakhsh et al. 2014). In addition, some clinicians have proposed the concept of maximum safe resection for GBM patients, which demands that neurosurgeons pursue the maximal extent of resection while Fig. 3 Nomogram predicted the 0.25 -, 0.5 -, and 1-years over survival for GBM patients presenting with distant extension 1 3 avoiding persistent neurologic deficit (Domino et al. 2020;Marko et al. 2014). As imaging diagnostics and intraoperative guidance technologies continue to improve, surgical decision-making should be made based on the pros and cons, rather than rigidly pursuing the extent of resection to improve survival rates, or adopting an overly conservative surgical approach due to poor prognosis. Our nomogram can potentially aid in the treatment decision-making process of GBM patients presenting with distant extension. Our research also has many limitations. Some important factors that may potentially affect prognosis were not documented in the SEER database, which include imaging data, clinical symptoms, results of immunohistochemistry and genetic testing of tumors, postoperative quality of life, etc. Due to the low probability of distant extension of GBM, the number of cases included in our research was relatively small. Finally, our study was retrospective, and we only performed internal validation. High-quality prospective studies are needed to validate our findings.

Conclusions
Stage (GBM patients with distant extension) is independent prognostic factor for GBM patients. Age, surgery, radiotherapy and chemotherapy are independent prognostic factors for GBM patients presenting with distant extension. Aggressive surgical strategy, beam radiotherapy, and chemotherapy can significantly improve the prognosis. The nomogram based on the above factors accurately predicted the 0.25-year, 0.5-year and 1-year OS of these patients.