Association Between the Extent of Resection and Prognosis in Adult Patients With WHO Grade III Gliomas: a Population-based Study

Dongjie He Air Force Medical University 2nd A liated Hospital: Air Force Medical University Tangdu Hospital https://orcid.org/0000-0002-4494-9950 Siying Zhu Air Force Medical University Peiwen Wu Air Force Medical University Qiming Wang Air Force Medical University Gaiyan Li Air Force Medical University Yuhong Qi Air Force Medical University Bing Zhang Air Force Medical University Hao Chang (  jacky9703@163.com ) Air Force Medical University https://orcid.org/0000-0002-4443-5167 Qiuju Shao Air Force Medical University

Most studies have reported the overall survival (OS) and predictors of survival in the anaplastic glioma population, of which several have revealed that surgery is associated with improved survival in WHO grade III gliomas [5][6][7][8]. The extent of resection for anaplastic gliomas has not been associated with survival [9,10]. Therefore, the impact of the extent of resection (EOR) on improving survival in patients with this tumor type is currently unclear.
Few large-sample studies on surgery to facilitate long-term survival of WHO grade III gliomas have been conducted to date. Even if studies reported survival in this population, the follow-up time was short or the impact of other causes of death in this population was not considered [7,11,12]. Therefore, exploring trends in patients' survival of these tumors is necessary, including OS and cause-speci c survival (CSS) [13,14].
To facilitate evidence-based clinical decisions regarding the use of surgery in anaplastic gliomas, we aimed to explore whether patients who underwent gross total resection (GTR) can gain survival bene ts compared with N-GTR using data from the Surveillance, Epidemiology, and End Results (SEER) database.

Seer Coding And Covariates Included
The following demographic data were obtained for analysis: age at diagnosis (patients were divided into groups of 20-49 and 50+ years of age), gender, race (white, black, other, and unknown), marital status (single, married, other, and unknown), year of diagnosis (2006-2010, 2011-2015, and 2016-2018), and tumor location (supratentorial, infratentorial, brain overlap, and unknown). Regarding treatment options, the extent of surgery (N-GTR [biopsy and subtotal resection], GTR), and therapies after surgery (radiation [RT], chemotherapy [CT], no, and yes) were included. In the current study, OS was de ned as the time from surgery to death from any cause, and CSS was de ned as the time from surgery to death from anaplastic glioma.

Statistical analysis
Categorical data were compared using the chi-squared test or Fisher's exact test, as appropriate. The Kaplan-Meier curve (K-M) and log-rank test were utilized to provide univariable OS visualization, and multivariable Cox regression analysis of OS and subgroup analysis was also performed for each variable.
Nelson-Aalen curve and Gray's test were utilized to provide univariable visualization of cause-speci c death (CSD) when considering competing events, and Fine and Gray's competing risk regression was used to assess cause-speci c survival.
All statistical analyses were performed using Free Statistics software version 1.3 (R Foundation for Statistical Computing, Version 3.3.2)[16]. Differences were considered statistically signi cant at p <0.05.

Baseline demographic and clinical information
A total of 3979 patients with anaplastic gliomas were included in the present analysis. Supplement.1 presents a ow chart of the participants. Among them, 2064 (51.87%) and 1915 (48.13%) underwent N-GTR and GTR, respectively. The proportion of patients under 50 years of age (54.71%) was higher than that over 50 years of age (45.29%). Factors associated with the extent of resection included age at diagnosis, sex, race, marital status, year of diagnosis, location, histologic type, radiation, and chemotherapy (Table 1). Most patients were white and married. The most common tumor location and histologic type was supratentorial (85.47%) and anaplastic astrocytoma (54.96%), respectively. Most postoperative patients received RT (76.28%) and CT (70.95%) in both arms. Overall Survival and cause speci c death of patients The impact of patient OS was graphically presented in the K-M survival curves. The median survival was 45 and 101 months for the N-GTR and GTR patients, respectively. The 5-year and 10-year OS in the GTR group were 59.9% and 45.0%, respectively, which were higher than the corresponding values of 44.0% and 29.4% in the N-GTR group, respectively (Supplement.2). The ve-year and ten-year OS of those who underwent N-GTR and those who underwent GTR differed signi cantly (P<0.001) (Fig. 1). The cumulative incidence curve of the GTR group is illustrated in Fig. 2, accounting for causes of death as a competing risk. The 5-year and 10-year cumulative incidence rates of cancer-speci c death in the N-GTR group were 51.9% and 65.5%, respectively, which were higher than the corresponding values of 36.6% and 49.9% in the GTR group, respectively (Supplement.3). The cumulative incidence function (CIF) was lower in patients who underwent GTR for cause-speci c death (P < 0.001), and did not differ signi cantly between the N-GTR and GTR groups for other causes of death (P = 0.689).
Multivariable analysis for OS and CSS among patients with grade III gliomas We employed multivariate Cox regression to adjust for confounding bias caused by unbalanced baseline variables and examine the prognostic effect of patients for OS and used competing risk regression to examine the prognostic effects of factors affecting CSS in our study (

Results of multivariate Cox analysis for overall survival in different subgroups
In the full cohort, GTR was associated with improved OS (HR: 0.72; 95% CI, 0.65-0.79; P<0.001) (Fig. 1,  Fig. 3). Considering that age, tumor location, histologic type, radiation, and CT may in uence the extent of resection in patients, we divided the cohort into ve subgroups to perform subgroup analysis (Fig. 3). Data showed that tumor location and histologic type played an interactive role in the association between the extent of resection and OS (both P for interaction< 0.001). Patients with an infratentorial phenotype exhibited lower OS, but the differences were not statistically signi cant (HR: 1.32, 95% CI, 0.75-2.32, P =0.332). However, subgroup analysis was performed according to the confounders including age, radiation, and chemotherapy, we did not found any signi cant interaction in these subgroups (P for interaction>0.05 for all).

Discussion
With the power of the SEER database, data from 3979 patients with anaplastic glioma were included, including 2064 who underwent either biopsy or partial excision and 1915 who underwent GTR, which enabled us to perform multivariable OS and CSS analyses. In the present study, we found a signi cant association between GTR and prolonged survival. We employed multivariate Cox regression and competing risk models to reduce confounding bias, and our ndings remained valid in these multivariate models.
The prognosis for anaplastic glioma remains poor despite combination treatment with surgery, RT, and CT [6]. In our cohort,48.13% of patients with anaplastic glioma underwent complete resection, and our ndings were consistent with those of previous studies, in which the percentage ranged from 30-60% in different reported cohorts [5,17,18]. Randomized controlled trials (RCTs) have shown that patients with anaplastic glioma are more likely to bene t from postoperative adjuvant treatment [19,20]. We also found that anaplastic astrocytomas accounted for approximately 54.96% of cases, making them the highest proportion among all patients with anaplastic glioma. Most postoperative patients included in our study had received radiation (76.28%) and CT (70.95%) in both arms. A possible explanation is that AA has a relatively high proportion, while adjuvant RT can improve survival [3], and CT prolongs survival for cases of AO or AOA [21,22].
Although improved survival has been correlated with greater extent of resection in anaplastic glioma [12,18,23,24], the longer survival bene t has not yet been determined. Other clinical studies have also suggested that patients with anaplastic glioma might not bene t from surgical resection [9,10]. Two measures of survival were calculated for each resection group, namely OS and CSS, as each of these measures has unique strengths [25]. Some studies analyzed 5-year OS instead of 10-year OS, and did not assess speci c survival. We found that the 5-year and 10-year OS rates of the GTR group (59.9% and 45.0%, respectively) were higher than those of the N-GRT group (44.0% and 29.4%, respectively) (p<0.001). This is consistent with the results of several other studies in which those who underwent GTR exhibited improved survival rates [5,18]. However, OS analyses after tumor diagnosis may be heavily in uenced by competing causes of death; therefore, the CSD rate was analyzed in the present study, taking deaths unrelated to anaplastic glioma and confounding bias into consideration. The 5-year and 10-year CSD rates of the GTR group were lower than those of the N-GRT group in our cohort (p<0.001), whereas the difference in death from other causes between the two groups was similar (p=0.689). After controlling for competitive risk [26], the extent of resection was not found to be associated with a decreased risk of CSD in patients with metastatic tumor. However, we found that the risk of CSD was higher in the N-GRT group than in the GTR group. We speculated that this might be related to the relatively long survival time of the patients in the present study. The results of our study demonstrated that GTR was associated with increased survival in these patients.
Several studies have analyzed the risk factors associated with survival among patients diagnosed with grade III gliomas, including age, sex, tumor location, histologic type, extent of resection, and postoperative adjuvant therapy [27] [28] [20,29,30]. However, few studies have reported the relationship between CSS and the extent of surgical resection in patients with anaplastic glioma. The GTR group was signi cantly associated with improved OS (HR: 0.72, 95% CI, 0.65-0.79, < 0.001) and CSS (HR: 0.72, 95% CI, 0.65-0.80, < 0.001) compared with the N-GTR group in the present study.
Interestingly, the association between OS and the extent of resection was stable for other clinical subgroups with respect to age, radiation, and CT. This showed that the extent of resection was positively associated with better OS in younger ( . However, the associations in GTR were not observed in the infratentorial site and histologic type compared with N-GTR (both p for interaction <0.001). A possible explanation is that the aim of the surgical approach should be limited to infratentorial tumors due to the high incidence of surgery-related neurological impairment [31]. The bene t of maximal surgery may be attenuated in patients in AO or AOA-relevant subgroups because of the chemosensitivity of the histologic type [32]. GTR had OS bene ts on cases of supratentorial tumors and AA subgroups [28]. The effect of surgical resection in patients with anaplastic glioma should be explored further.
This study had several limitations. First, information on tumor molecular data or patient functional performance, which are factors that affect survival, were missing from the SEER database. Second, the detailed CT regimen and RT dosage were not recorded in the SEER database. Third, EOR threshold values are important for anaplastic gliomas [33]. Fourth, the SEER database does not provide information on disease recurrence or subsequent treatment data.

Conclusion
In summary, the current study suggested that GTR could provide an OS and CSS advantage in cases of grade III glioma, particularly in patients with supratentorial tumors or anaplastic astrocytomas. This study provides valuable data for further clinical studies and highlights the need for more clinical studies on histologically diagnosed grade III gliomas.

Statements And Declarations
Funding The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Competing Interests
The author(s) declared no potential con icts of interest with respect to the research, authorship, and no relevant nancial or non-nancial interests.

Author contributions
Dongjie He and Siying Zhu analyzed the data and drafted the paper. QiMing Wang, Gaiyan Li and Bing Zhang collected the data, QiMing Wang and YuHong Qi analyzed the data. QiuJu Shao and Hao Chang conceived and designed the idea to this paper. All authors drafted the work/revised, provided nal approval of the version, and agree to be accountable for all aspects of the study in ensuring that questions related to the accuracy or integrity of any part of the study.

Data Availability
The data were abstracted from an open database, the Surveillance,Epidemiology, and End Results (SEER) 18 Registries Data (https://seer.cancer.gov/). Researchers should use the link to request access to study data.

Ethics approval
This was a retrospective study. Therefore, the requirement for ethical approval and informed consent was waived for the present study.