Age is associated with unfavorable neuropathological and radiological features and poor outcome in patients with WHO grade 2 and 3 gliomas

With the rising life expectancy and availability of neuroimaging, increased number of older patients will present with diffuse and anaplastic gliomas. The aim of our study was therefore to investigate age-related prognostic clinical, neuropathological and radiological features of lower-grade gliomas. All consecutive patients with diffuse or anaplastic glioma WHO grade 2 and 3 who underwent first tumor resection between 2010 and 2018, were selected from the institutional neuro-oncological database and evaluated. The mean age of 55 males and 44 females was 46 years (SD ± 16). Wild-type IDH1 (p = 0.012), persistent nuclear ATRX expression (p = 0.012) and anaplasia (p < 0.001) were significantly associated with higher age. The CE volume before resection was found to be increased in older patients (r = 0.42, p < 0.0001), and CE rate was higher in the IDH wild-type population only (p = 0.02). The extent of resection did not differ with age. Overall, one year of life resulted in a PFS reduction of 9 days (p = 0.047); in IDH sub-group analysis, this dependency was confirmed only in wild-type tumors (p = 0.05). OS was significantly reduced in older patients (p = 0.033). In conclusion, behavior and prognosis of WHO grade 2 and 3 glioma were unfavorable in correlation to patient’s age, even if the extent of resection was comparable. Older age imparted a poorer PFS and higher CE rate only in the IDH wild-type population.


Materials and methods
All consecutive adult patients with neuropathologically proved intracranial WHO grade 2 and 3 gliomas, operated in our center for their first resection between January 2010 and March 2018, were selected from our prospective neurooncological database. The study was approved by the ethics committee of Medical University Innsbruck (AN5220 329/4.4) and written informed content was acquired from all participants. This study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.
Preoperative and postoperative MRI within 72 h were performed including T1-weighted Gadolinum-contrasted as well as native T1, T2, FLAIR and DWI sequences as the standard of care for patients harboring intracranial tumors 19 . The metric volume of alteration was manually measured using segmentation in ITK-SNAP software (v.3.8.0 for Mac OS, UPenn and UNC dev., http:// www. itksn ap. org) in T1 with contrast enhancement (CE) as well as native T1, T2, FLAIR and DWI sequences 20,21 .
Neuropathological examination was routinely performed on FFP-embedded tissue. Integrated neuropathological reports were based on the WHO grading system 22,23 . Presence of IDH1 mutation in the R132H position was assessed with IHC and, in case of negative result, DNA sequencing was performed for patients under 40 years to approve the IDH1 wild-type status. Nuclear ATRX and EGFR expression were tested with IHC. In case of lost ATRX, the oligodendroglial genotype was proved with the unbalanced co-deletion of 1p/19q chromosomal regions through Fluorescence in situ hybridization (FISH).
After the final neuropathological conclusion, each case was individually discussed in the institutional multidisciplinary tumor-board to establish the adjuvant management. The standardized recommendation was based on international guidelines 10,11,24,25 . In case of higher risk tumors -anaplastic glioma, incomplete resection, wild-type IDH1 or lost nuclear ATRX -the radio-chemotherapy according to previously published protocols 11,26 was applied; and if the mentioned higher risk tumor criteria were absent, the wait and see strategy was chosen.
Statistical analysis and graphics generation was processed using IBM SPSS Statistics (IBM SPSS Statistics for Mac OS, Version 26.0. Armonk, NY: IBM Corp., https:// www. ibm. com/ analy tics/ spss-stati stics-softw are). For analysis, patients were divided in two groups according to the median age: younger than median and older than median. Normal distribution of scale parameters was checked by the Kolmogorov-Smirnov test. Correlations for non-parametric data were determined using the Spearman's method. T-Tests for normal distributed scale parameters, Mann-Whitney U-test for rank and scale parameters lacking normal distribution, and Chi 2 -test comparing two binominal parameters were applied according to general terms of these tests. The linear and Cox regressions were used to reveal the progression free survival (PFS) and overall survival (OS) dependencies. The Holm-Bonferroni (H-B) correction was corrected for multiple hypothesis 27,28 . The confidence interval and α were defined as 95%.
Ethics approval. The study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. The database and tissue bank are approved by the ethics committee of Medical University Innsbruck (AN5220 329/4.4).

Consent to publication.
No individual data is showed separately in the manuscript. All data is used only after anonymized statistical processing and is described with pool results.

Results
In total, 99 patients were eligible for the analysis: 55 (56%) males and 44 (44%) females with a mean age of 46 years (SD ± 16; range 18-85). Patients harboring more aggressive tumors were older: 47 patients with diffuse glioma had a mean age of 40 years (SD ± 14) compared to 51 years (SD ± 16) in 52 patients with anaplastic transformed gliomas (p < 0.001, H-B: 0.003). IDH1 was mutated in 71% and wild-type in 29% of cases. ATRX was expressed in 62% and lost in 38% of LGG samples.
CE was present in 26% (13/50 of < 45y.) of cases in younger group and in 63% (31/49 of ≥ 45y.) of cases in older group (p < 0.001). Patients with higher age, who had IDH wild-type glioma, showed CE more frequently (p = 0.02); on the other hand, we could not find analogous dependency in case of IDH-mutated tumors (p = 0.11). CE volume before resection directly correlated with age, while the native volumes did not show differences (Table 1).
There were no differences in the extent of resection in relation to age. Complete resection of contrast enhancing tumor parts, if present, was achieved in 88% compared to 74% (younger vs. older; p > 0.05). Thus, 96% patients of younger group and 93% of older group did not have any CE after resection. The volume of residual CE tumor, however, correlated with age; r = 0.27, p = 0.013. The probability of gross total resection for FLAIR (83% vs. 78%) and DWI (85% vs. 78%) was similar between our age-related groups, p > 0.05.

Discussion
Older patients with diffuse and anaplastic WHO grade 2 and 3 gliomas had higher CE volume and more frequently harbored negative prognostic molecular markers like wild-type IDH. Although resection rate was similar, the residual CE volume was higher in older population. Overall, one year of patient's life reduced the PFS by 9 days and OS was significantly decreased in relation to age. Moreover, in IDH sub-group analysis, age-dependent PFS reduction as well as higher CE rate in MRI was confirmed only in case of wild-type tumor.
Age is related with unfavorable neuropathological prognostic factors in case of LGG. The higher incidence of anaplasia in older patients results in worse PFS 26,29 . Patients with IDH wild-type and retained nuclear ATRX expression are significantly older, which is concordant and well-known according to the literature [30][31][32][33] and this neuropathological profile is associated with aggressive behavior [34][35][36][37][38] . Similar findings were reported for glioblastoma, where older patients showed less frequently IDH mutation and consequently worse clinical outcome 39,40 . Moreover, this data is concordant to the fact, that patients with IDH wild-type LGG have an increased risk of malignant transformation 41,42 .
Positive CE in MRI is usually interpreted as prognostically unfavorable feature 43,44 . We showed, that agedependent higher CE rates occur only in case of IDH wild-type glioma: that is a further evidence of the added aggressiveness of these tumors in older population. Thus, it is possible, that the relationship between age and CE is driven by the underlying relationship between age and IDH status.
If a patient was eligible for surgery, it was possible to provide the same extent of resection independently from age. Perioperative complication rates did not differ within elderly people as reported before 45 . Thus, neurooncological surgeons should aim for a radical resection for these patients as well 9 , as it provides more beneficial outcome according the actual guidelines 9,10 and comparative studies 46 . However, the hypothesis that older patients with LGG are undertreated has already been suggested in other studies 47 . According to routine clinical standards, patients are not undertreated in our institution only due to their higher age.
The prognostic unfavorable molecular features consequently led to decreased PFS. We were able to show a predictable association between age and PFS, which could be a great help for physicians in estimating of the expectancies and in treatment suggesting: according to our data one year of life shortens the PFS by 9 days. Thus, a patient with 40 years of age had one year more calculated PFS compared to a patient twice that age. Furthermore, this association was present only in case of IDH wild-type gliomas according to sub-group analysis. Thus, even if there is dependency between older age and less favorable molecular characteristics, older age still imparts a poor prognosis in the IDH wild-type population. The worse PFS corresponded to a significantly worse OS of the older patients as well.
Our study has limitations. The survival endpoints were not reached for all patients, which could limit Cox regression analysis of OS in IDH sub-groups. We did not consider the pre-existing diseases, which were not recorded in our prospective database and could not be gained retrospectively keeping acceptable quality of data. However, patients were processed in our study only if they underwent substantial resection. That includes only patients who were eligible for intracranial surgery with or without general anesthesia. Thus, critically ill patients were indirectly excluded and did not influence our results. The sub-group analysis was restricted in our study due to limited number of cases. Survival analysis requires external validation.

Conclusions
Age truly matters in LGG. Distinct tumor features like wild-type IDH, malignant behavior and dismal prognosis were more unfavorable in older population, even if the extent of resection was similar. We were able to compute a PFS reduction for 9 days with each year of patient's life. However, greater age imparted a poorer PFS and higher CE rate only in the IDH wild-type population. That needs to be evaluated in a large prospective cohort.

Data availability
The raw data was generated in authors' institution. The data that support the findings of this study are available on reasonable request from the corresponding author. The data are not publicly available due their containing information that could compromise the privacy of research participants. www.nature.com/scientificreports/