Survival outcome and prognostic factors in anaplastic oligodendroglioma: a single-institution study of 95 cases

The aim of this study was to evaluate prognostic factors including surgical, radiographic, and histopathologic analyses in anaplastic oligodendroglioma (AO) patients. We reviewed the electronic records of 95 patients who underwent surgery and were diagnosed with AO for 20 years. The primary endpoints were progression-free survival (PFS) and overall survival (OS). Univariate and multivariable analyses included clinical, histopathological, and radiographic prognostic factors. Subgroup analysis was performed in isocitrate dehydrogenase (IDH1/2)-mutant and 1p/19q-codeleted patients. The median PFS and OS were 24.7 months and 50.8 months, respectively. The 1-, 3-, 5-, and 10-year PFS were 75.8%, 42.9%, 32.4%, and 16.4%, respectively. Furthermore, the 1-, 3-, 5-, and 10-year OS were 98.9%, 76.9%, 42.9%, and 29.7%, respectively. The median PFS and OS of the IDH1/2-mutant and 1p/19q-codeleted patients were 54.2 and 57.8 months, respectively. In univariate analyses, young age, frontal lobe, weak enhancement, gross total resection (GTR), low Ki-67 index, 1p/19q codeletion, and IDH1/2 mutations were associated with a favorable outcome. In multivariable analyses, IDH1/2 mutation was related to better PFS and OS. In subgroup analysis, GTR was associated with favorable outcomes.

Anaplastic oligodendroglioma (AO) is a rare disease entity, comprising 0.5% of all intracranial neoplasms 1 . The current standard for AO treatment consists of maximum safe resection and radiotherapy (RT) followed by chemotherapy (CTx) 2 .
Since the 2016 World Health Organization (WHO) classification of glioma, molecular studies and chromosomal analysis have been deemed essential for diagnosis. Mutations in the isocitrate dehydrogenase genes (IDH1/2) and chromosome 1p/19q codeletion status are key in defining oligodendroglioma. Most previous research usually examined oligodendroglial tumors, which include both oligodendroglioma (OD) and oligoastrocytoma (OA) [3][4][5] . Hence, research associated with only AO is very rare. Liu et al. reported that 5-year and 10-year survival rate of AO was 50.2% and 36.2%, respectively. They included 1899 patients who had a histological diagnosis only 6 . Kang et al. published a multicenter study of AO which included 376 patients from nine Korean institutes. They reported that 5-year and 10-year survival rate of AO was 58% and 45%, respectively. The median PFS was 34.5 months, and the actuarial 5-and 10-years PFS rates were 38% and 24%, respectively 7 . The limitation of those studies was histological diagnosis, not molecular diagnosis. Our study tries to overcome this limitation by subgroup analysis which included molecular diagnosis as IDH1/2-mutant and 1p/19q codeleted AO.
The aim of this study was to evaluate prognostic variables, including extent of surgical resection (EOR), radiographic features, and molecular markers, in a large single-center series of AO patients. In addition, according to these results, we performed a subgroup analysis in IDH1/2-mutant and 1p/19q-codeleted AO patients to predict their prognosis.
1. Underwent surgery for the first time for anaplastic oligodendroglioma 2. Age at surgical procedure ≥ 20 years old 3. Followed for more than 1 year after surgery 4. EOR evaluated by pre-and postoperative MRI 5. No duplicated or recurrent cases Study design. The primary endpoints of this study were progression-free survival (PFS) and overall survival (OS). The date of diagnosis was defined as the date of initial surgery and used as the starting date for PFS and OS, with survival measured until either tumor growth was noted on imaging follow-up, latest follow-up, or death.
We performed clinical analyses in 95 AO patients to determine predictive factors for PFS and OS. Subsequently, subgroup analysis was done for patients who had both IDH1/2-mutant and 1p/19q codeletion, which is the current molecular definition of AO.
Extent of resection and adjuvant treatment. EOR was assessed by comparing preoperative CT and MRI with postoperative CT and MRI, which was usually taken in the 48 h following surgery. Gross total resection (GTR) was defined as the removal of more than 99% of enhancing lesions 8 .
Adjuvant therapy comprises RT, CTx in the form of either PCV (procarbazine, lomustine, and vincristine) or temozolomide, or RT followed by CTx. Chemotherapeutic regimens followed the standard of treatment 9 . RT was performed with three-dimensional (3D) conformal RT or intensity-modulated RT. The RT dose was 60 Gy over 30 fractions in most cases. The RT volume was the surgical bed and T2 (or FLAIR) hyperintensity on postoperative MRI plus a margin of 1.5-2 cm. Radiographic evaluation. Periventricular lesions (ie, subventricular zone involvement) were defined as any T2-weighted, FLAIR, or T1-Gd-enhanced portion contacting any ventricular system. Calcification was assessed according to preoperative CT. Enhancement patterns were classified into two categories: strong or weak (Fig. 1). Strong enhancement was defined by rim enhancement, whereas weak enhancement was patchy, focal, and faint enhancement. Tumor progression was assessed on follow-up MRI according to The Response Assessment in Neuro-Oncology (RANO) criteria for gliomas 10 . When in doubt, our standard of practice was to use MR spectroscopy and positron emission tomography.
Histopathologic factors. The Ki-67 labeling index was defined as the percentage of cells with positive nuclear stain. The fraction of labeled nuclei was determined by eye using a grid and expressed as a percentage of five intervals. Oligodendroglial tumors have been routinely tested for IDH1 mutation status using immunohistochemistry (IDH1 R132H, MPQ-67, 1:50 dilution; Cell Marque, Rocklin, CA) since 2012 in our institute. Subsequently, DNA sequencing was performed for negative IDH1 immunohistochemistry. Exon 4 of the IDH1 and IDH2 genes was amplified by polymerase chain reaction and sequenced from tumor samples, as done by  Consent to participate. Because of its retrospective manner, informed consent was waived.
Thirty-six (37.8%) patients received adjuvant RT without CTx whereas five (5.2%) patients were given CTx without RT. Among 36 patients who received RT but not CTx, they were included in KNOG-1101 study for RT single arm or had co-morbidities. Forty-eight (50.5%) patients had RT followed by CTx. Specifically, 27 of 48 (56.3%) patients had temozolomide and 21 (43.8%) had PCV chemotherapy. Chemotherapy regimens were decided by patients themselves based on socioeconomic status, comorbidities, possibilities of adverse effects, and compliance. Univariate analysis (log-rank test) based on adjuvant treatment methods showed that the adjuvant RT/PCV group had a better PFS than those with adjuvant RT followed by temozolomide (median PFS 32.2 vs. 24.7), but this was not significant (p = 0.621). The median OS for each group was 57.6 months and 57.8 months, respectively (p = 0.573).  www.nature.com/scientificreports/ Multivariable analysis. Figure 3 and Table 2   Based on our results, we should attempt GTR which could be enhanced by 5-aminolevulinic acid (5-ALA) 16 . PCV chemotherapy has been the standard chemotherapeutic regimen for grade 2 or 3 glioma patients 14,18 . In contrast, temozolomide has been utilized as an adjuvant treatment for AO patients 18 . Our study revealed that RT followed by PCV chemotherapy yielded a longer median PFS (32.3 vs. 24.7 months, p = 0.6209) than RT with temozolomide; however, OS was not significantly different between these groups (57.6 vs. 57.8 months, p = 0.5729). PCV and temozolomide have their own advantages and disadvantages for patients. Our results support the notion that temozolomide is not inferior to PCV in relation to OS, although this is cannot be concluded with certainty as this was a retrospective study.

Radiographic evaluation.
It is difficult to differentiate AO from OD (WHO grade II) based on CT or MRI, as OD and AO share radiographic findings including calcification and enhancement. There are slight but unspecific differences in MRI. OD also shows minimal or moderate enhancement in many cases 19 . However, strong enhancement or rim enhancement patterns are rare. Additionally, the incidence of calcification in our AO patients was relatively lower than in OD patients (90% previously reported 19 vs. 36% in our population).
Prognosis related to enhancement pattern has been evaluated in anaplastic astrocytomas in several studies 20 . However, this is the first study to examine prognosis and enhancement patterns of only AO, although our results did not reach statistical significance in multivariable analysis. Based on our study, we predict relatively earlier disease progression after surgery in patients with strong enhancement.
Periventricular involvement has been regarded as a poor prognostic factor in GBM patients 21 . Lim et al. revealed that a GBM involving neural stem cell regions (like the subventricular zone) tends to have multifocal lesions and a greater rate of recurrence, resulting in poorer outcomes. However, in the present study, we could not find a relationship between periventricular involvement and survival outcome.
Ki-67 index. The Ki-67 index has been evaluated in several studies related to AO. Preusser et al. reported that a low Ki-67 index was associated with significantly longer PFS and OS in univariate analysis 4 . Furthermore, Celso et al. revealed that high Ki-67 correlated with shorter OS, both in univariate and multivariable analyses 22 . In our study, a lower Ki-67 index was related to a better survival outcome (HR 0.16, 95% CI 0.28-0.87, p = 0.034). IDH1/2-mutant and 1p/19q-codeleted AO patients. 1p/19q codeletion and IDH1/2 mutation are well-known favorable factors in OD patients 23 . After the introduction of the 2016 WHO classification, IDH1/2 mutation status was key for classifying glioma patients. In our institute, IDH1/2 mutation status of glioma patients has been routinely available since June 2012. Other patients without data on IDH1/2 mutation or with IDH1/2-wild type AO were diagnosed with AO, NOS. Clinically, there are many AO patients who have been www.nature.com/scientificreports/ followed up without molecular analysis results. To include patients with unknown IDH1/2 mutation status but diagnosed with AO before the 2016 WHO classification, our study included all AO patients before and after 2016. We tried to identify predictive factors in AO patients. Subsequently, we were expecting to reach a similar conclusion in IDH1/2-mutant and 1p/19q-codeleted AO patients. Fortunately, among the variables which were meaningful in AO patients, GTR was associated with PFS in univariate analysis for IDH1/2-mutant and 1p/19qcodeleted AO patients. Due to the small number of patients, we were unable to make a strong conclusion in our study. However, this result can give direction for further study. Further larger studies (or meta-analyses) should consider the factors mentioned in this study (enhancement pattern, Ki-67 index, and EOR) and determine their exact prognostic value.

Conclusion
Age under 60 and lower Ki-67 index were associated with longer OS. In subgroup analysis of IDH1/2-mutant and 1p/19-codeleted AO patients, the GTR group had better PFS than the non-GTR group. We recommend further studies including our factors to determine associations between PFS or OS in IDH1/2-mutant and 1p/19qcodeleted AO patients.