Introduction

Primary systemic chemotherapy, i.e. upfront chemotherapy, after surgical removal is being increasingly used as a treatment option not only for oligodendroglial tumors but also for low-grade astrocytic tumors [1–4]. Although histological distributions and the ratios of enhancing tumors vary between studies, the response rates of low-grade gliomas (LGGs) to either temozolomide or PCV (procarbazine, lomustine, vincristine) chemotherapy are reportedly between 30 and 75 % [2, 5–10]. Recent advance in translational research has provided molecular correlates with response to chemotherapy, and 1p19q codeletion, mutations in isocitrate dehydrogenase (IDH)1/2 genes, and possibly promoter methylation of the MGMT gene are predictive of chemotherapeutic response in grade II or III gliomas [7, 8, 11–14].

Neoadjuvant chemotherapy is a common treatment strategy for many cancers [15, 16]. The primary aim of this strategy is to reduce the size or extent of cancer before a main treatment, and thus decreasing the difficulty and morbidity of procedures with improved functional and/or survival outcomes. Although functional preservation is critical in glioma surgery, the neoadjuvant strategy has not been systematically used thus far.

Since February 2006, we have tailored treatment for diffuse gliomas according to molecular information. Specifically, after maximal safe removal, diffuse gliomas of grade II or III with favorable molecular status for chemosensitivity were treated with upfront chemotherapy regardless of histological diagnosis. In cases of incomplete initial surgery, a second surgery was advised after reducing tumor volume by chemotherapy. Because of invasion into eloquent areas, initial surgery was incomplete for the majority of the cases in the series, providing a unique opportunity to assess chemotherapeutic response and the utility of the neoadjuvant strategy in the consecutive cases selected by molecular profile.

Materials and methods

Study design and patient population

This is a prospective, observational study of the consecutive patients treated at Keio University Hospital between 2006 and 2013 who fulfilled the following criteria: (a) age > 18 years, (b) supratentorial tumor location, (c) histopathological diagnosis of diffuse glioma of WHO grade II or III, (d) no prior chemotherapy or radiotherapy, (e) 1p19q codeletion or promoter methylation of the MGMT gene, (f) written informed consent for molecular analysis of the tumor and for translational research approved by the Institutional Review Board at Keio University (Approval Number 20050002).

Treatment

The initial surgery was performed using intraoperative mapping technique including awake craniotomy with the aim of more than subtotal removal of abnormal high intensity area on T2-weighted or fluid-attenuated inversion recovery (FLAIR) MRI, or, as the second best, removal of increased methionine-uptake on positron emission tomography with 11C-methionine (11C-methionine PET) (Online Resource 1). For the cases in which these radical resections were considered high risk, biopsy or debulking surgery only to reduce mass effect was performed. After initial surgery, the patients with potentially chemosensitive gliomas based on molecular status were treated with upfront chemotherapy either by temozolomide, ACNU monotherapy, or PAV combination chemotherapy. Temozolomide was orally administered between days 1 and 5 in a 28-day cycle at a dose of 150–200 mg/m2/day. ACNU monotherapy was administered every 6–8 weeks at a dose of 80 mg/m2/day on day 1. PAV chemotherapy was administered every 8 weeks as follows: 70 mg/m2 intravenous nimustine on day 1, 60 mg/m2 oral procarbazine on days 8–21, and 1.4 mg/m2 vincristine (maximum dose, 2 mg/body) intravenously on days 8 and 29. The patients chose chemotherapy at their discretion after considering its potential efficacy, toxicity, and cost, according to information given by their primary neuro-oncologist (HS or TY). The upfront chemotherapy was given at progression in several patients who preferred watchful wait to immediate chemotherapy following initial resection. In cases of incomplete initial resection, second-look radical resection was intended after maximal response to chemotherapy, and grade III tumors were irradiated after chemotherapy or second-look resection. Grade II tumors, including those with focal anaplastic changes, were observed after chemotherapy or second-look resection. The present treatment strategy was approved by the Institutional Review Board at Keio University (Approval Number 20130250). Radiotherapy was administered at the time of recurrence in the majority of recurrent cases.

Neuroimaging

Imaging follow-up was performed by MRI (1.5 or 3T) every 3 cycles of temozolomide or 1–2 courses of nitrosourea-based chemotherapy. Tumor volume was assessed on T2-weighted or FLAIR MRI, and maximal chemotherapeutic response was defined in the absence of signs of improvement on next MRI. Response assessment in neuro-oncology (RANO) criteria was used with some modification, and percent change in tumor volume following chemotherapy was calculated by the sum of the product of perpendicular diameters (SPD) [17]. In cases of slowly growing non-enhancing areas that were indistinguishable from treatment-related changes, the date of progression was defined by any of the following: (1) unequivocal extension of the non-enhancing area, (2) development of enhancement, 3) clinical deterioration, or (4) increased methionine-uptake on 11C-methionine PET. Stable disease (SD) was defined as any other clinical status that does not meet the criteria for complete response (CR), partial response (PR), minor response (MR), or progressive disease (PD), lasting at least 6 months [8, 17, 18]. In selected cases, 11C-methionine PET was performed once a year or when tumor recurrence was doubted on MRI.

Histopathological and molecular-genetic examinations

Histopathological diagnoses were made according to the WHO classification [18]. The codeletion of 1p19q was assessed using comparative genomic hybridization [19, 20], promoter methylation of the MGMT gene was assessed using methylation-specific PCR [21], and mutations of IDH1/2 genes were assessed using immunohistochemistry with an anti-mutant IDH1 antibody (Dianova, Hamburg, Germany) or by sequencing [22, 23].

Statistical analysis

Progression-free survival (PFS) and overall survival (OS) were calculated from the date of initiation of upfront chemotherapy. Survival curves were estimated using the Kaplan–Meier method with JMP version 8.0.1 (SAS Institute, Cary, NC).

Results

Clinical characteristics

Among 26 patients who fulfilled the inclusion criteria, 22 had grade II tumors and 4 had grade III tumors (Online Resource 2, Table 1). Nineteen out of the 26 cases were newly diagnosed, and seven were progressive disease after wait and see following initial surgery. Presenting symptoms included seizures in 15 patients, subjective symptoms such as headache and vertigo in 3, and focal symptoms such as aphasia, hemianopsia, and apraxia in 4. The other 4 tumors were found either during medical check-ups or incidentally. Because most of the patients with gliomas referred to our hospital had invasions into eloquent areas, initial surgery was incomplete in the majority of cases, with biopsy in 11, partial removal in 8, subtotal removal (STR, > 95 %) in 4, and gross total removal (GTR, total removal of T2/FLAIR abnormality) in 3.

Molecular diagnoses and upfront chemotherapy

Clinical and molecular characteristics of the study population are presented in Table 1 (Online Resource 2). These 26 tumors were deemed chemosensitive according to the presence of 1p19q codeletion and/or MGMT promoter methylation. Seventeen tumors showed both 1p19q codeletion and MGMT methylation, 5 showed codeletion and unmethylated MGMT, 2 had codeletion with an unknown MGMT status, and 2 had no codeletion with methylated MGMT. Mutations of IDH genes were found in all 24 tumors that were successfully analyzed, including 22 with IDH1 mutations and 2 with IDH2 mutations. Upfront chemotherapy was immediately administered after initial surgery in 19 cases and at tumor progression following the observation period in 7 cases. Chemotherapy regimens comprised temozolomide in 20 patients and nitrosourea-based chemotherapy in 6. The median number of courses of temozolomide and nitorosourea-based chemotherapy was 12.5 (range, 6–40) and 3.5 (range, 3–6), respectively. One of the primary neuro-oncologists preferred long prescriptions of temozolomide. The regimens were well tolerated, although a patient (case 22) suffered from grade 4 myelosuppression after the 4th course of PAV therapy, and another patient (case 13) suffered from liver dysfunction after the 1st course of PAV therapy and was subsequently administered ACNU monotherapy. In the majority of cases, treatment was discontinued following 3–6 (for temozolomide) or 1–2 (for nitrosourea-based chemotherapy) additional courses after maximal response. Maximal tumor responses to chemotherapy were noted after a median of 6 courses of temozolomide (3–15) or 3 courses of nitrosourea-based chemotherapy (2–4; Table 1, Online Resource 2).

Chemotherapeutic response and second-look resection following tumor volume reduction

Of the 22 evaluable cases, objective response (CR + PR + MR) was noted in 14 (64 %), and no PD was observed within 6 months of chemotherapy initiation (Table 1, Online Resource 2). In 21 out of the 22 cases, percent change in tumor volume was calculated using SPD. The median SPD change following chemotherapy was −35 % (−100 to +2.1), and all but one case showed volume decrease after upfront chemotherapy, as estimated according to T2/FLAIR abnormality (Fig. 1). No clear differences in percentage tumor volume decrease were observed between non-codeleted or MGMT-unmethylated cases and cases with both codeletion and methylation (Fig. 1). Importantly, upfront chemotherapy separated T2/FLAIR abnormality from vital structures in several cases. Moreover, in some other cases, area of methionine uptake contracted to resectable size, despite that T2/FLAIR abnormality remained unresectable. These reduction in T2/FLAIR volume or methionine uptake enabled second-look radical resection, which was considered high risk at presentation or before chemotherapy. As a result, the second-look resection following tumor volume decrease was performed in 13 out of the 22 evaluable cases, i.e., in 12 out of the 19 cases with incomplete initial surgery: GTR/STR in 10 cases and removal of the area of methionine uptake in 3 [Fig. 1; Table (Online Resource 2)].

Fig. 1
figure 1

Maximum decrease of tumor volume (T2/FLAIR) from the baseline according to the sum of the products of tumor diameters (SPD) in 21 evaluable cases. The vertical axis indicates percent changes from the baseline to the minimum SPD after the beginning of upfront chemotherapy. Patients are indicated on the horizontal axis. Negative values in the vertical axis indicate tumor volume decrease from the baseline; all but one patient showed tumor volume decrease. Slashed boxes indicate cases that underwent second-look resection after upfront chemotherapy. Arrow codeleted but MGMT unmethylated case. Arrow head: MGMT methylated but non-codeleted case

Full size image

Case illustration

Case 1 (Fig. 2)

A 52-year-old man was diagnosed with a tumor in the right motor cortex. Biopsy revealed oligodendroglioma (1p19q codeletion, MGMT methylation, IDH1 mutation). Upfront temozolomide (14 courses in total) contributed to separation from the pyramidal tract and hand motor cortex, resulting in restriction of the tumor to the face motor cortex (PR, 56 % decrease by SPD). Subsequently, GTR was performed with an awake procedure, resulting in the cessation of seizures and no evidence of recurrence for 5 years and 5 months (5y 5 m) after the initiation of the upfront chemotherapy (4 years 3 months after second-look resection; KPS 100). Radiotherapy was not performed.

Fig. 2
figure 2

Serial MR images (FLAIR) of a case of second-look resection (case 1) a–c, e–g, i–k axial images. d, h Coronal images. a–c Before chemotherapy (before biopsy). The tumor was mainly located in the right precentral gyrus with invasion into the hand motor cortex and corona radiata. d–g After 12 courses of temozolomide. SPD of the tumor was decreased by 56 %. Upfront temozolomide separated the tumor from the corona radiata, and the tumor became restricted within the face area of the motor cortex. h–k After second-look removal (1 POD). Gross total removal of FLAIR abnormality was achieved

Full size image

Case 7 (Fig. 3)

A 43-year-old man had initially undergone a partial removal of his right frontal tumor, and histopathological diagnosis was oligodendroglioma (1p19q codeletion, MGMT methylation, IDH1 mutation). After observation, the tumor progressed (5 years 6 months after the initial operation). Because he suffered from depression at that time, extensive resection of the frontal lobe was evaded, and 3 courses of upfront PAV were prescribed. The tumor showed a partial response (61 % decrease by SPD), and tumor invasion into the insular cortex disappeared. Subsequently, STR of the T2/FLAIR abnormality was performed with the indication that the frontal margin was incomplete. Tumor recurrence was noted at the frontal area 4 years 9 months after the initiation of PAV (2 years 8 months after second-look resection). However, tumor recurrence was not demonstrated in the insular cortex, where the T2/FLAIR abnormality was initially observed. The patient subsequently underwent right frontal lobectomy and radiotherapy, and is alive and tumor free at 7 years 5 months after the initiation of PAV (KPS 90).

Fig. 3
figure 3

Serial MR images (FLAIR) of a case of second-look resection (case 7). a–c At progression. The histopathological diagnosis of the initial resection was oligodendroglioma. A diffuse right frontal tumor is shown invading the insular cortex; d–f after 3 courses of PAV. SPD of the tumor was decreased by 61 %. No tumor invasion into the insular cortex is apparent; g–i after second-look removal (1 POD). Subtotal removal was achieved with some residual FLAIR abnormality at the anteromedial margin (arrow); j–l at recurrence (4 years and 9 months after the initiation of PAV). Despite diffuse tumor recurrence in the right frontal area, there is no recurrence in the insular area where the FLAIR abnormality was initially shown (arrowhead)

Full size image

Case 26 (Fig. 4)

A 56-year-old woman had initially undergone a partial removal of her right frontal tumor, and histopathological diagnosis was diffuse astrocytoma (non-codeletion, MGMT methylation, IDH1 mutation). After observation (8years 7 months after initial operation), the tumor progressed and was treated with upfront temozolomide (15 courses in total). The tumor showed MR (30 % decrease by SPD), and partial frontal lobectomy, including the residual methionine-uptake area, was performed. The patient is free of recurrence at 5years 6 months after the initiation of upfront TMZ (4 years after second-look resection; KPS100). Radiotherapy was not performed.

Fig. 4
figure 4

Serial MR images (FLAIR) and 11C-methionine PET images of a case that underwent second-look resection (case 26) a–d at progression; the histopathological diagnosis of the initial resection was diffuse astrocytoma. After observation, a diffuse right frontal tumor invading the basal ganglia is shown; e–i after 15 courses of temozolomide. SPD of the tumor was decreased by 30 %. On the PET/CT image (e), some residual methionine uptake is shown in the right frontal lobe; j–n after second-look resection (2 months after resection). Partial frontal lobectomy, including the methionine uptake area, was performed

Full size image

Histopathological diagnosis and molecular characteristics of the second-look resected tumors

The histopathological diagnosis of the tumors resected following tumor volume decrease was similar to that of the initial tumors, except for a progressive oligodendroglioma with malignant transformation (case 8; Table 2, Online Resource 2). Proliferative indices (MIB-1, Dako, Glostrup, Denmark) of the second-look resected tumors were generally lower than those of the initial tumors, except for case 8 (Table 2, Online Resource 2), suggesting that the second resection was performed under chemotherapeutic control. The promoter methylation status of the MGMT gene in the second-look resected tumors was unchanged as compared to those in the initial tumors.

Survival analysis

During the median 57-month follow-up period (12–89 months), recurrence was noted in 12 out of the 26 cases, and 4 patients died. The median PFS among the 22 patients with grade II gliomas who were treated with the neoadjuvant strategy, including those with focal anaplasia, was 57 months, and OS remained less than the median (Fig. 5).

Fig. 5
figure 5

Progression-free survival (PFS, a) and overall survival (OS, b) of the 22 patients with grade II gliomas (20 codel, 2 non-codel) treated with the neoadjuvant strategy. PFS and OS were calculated from the beginning of upfront chemotherapy

Full size image

Pattern of recurrence after second-look resection

Six out of the 13 patients who underwent second-look resection showed recurrence afterwards. In all but one case, tumor regrowth occurred around incomplete surgical margin based on T2 or FLAIR MRI (case 7; Fig. 3). In case 23, tumor dissemination on the contralateral ventricular wall was missed before upfront chemotherapy, and recurrence occurred there despite no evidence of recurrence around the surgical cavity of the second-look resection. Case 18 showed an aggressive clinical course after recurrence at the incomplete surgical margin.

Discussion

Some recent studies suggested the efficacy of neoadjuvant chemotherapy for gliomas of WHO grades II–III [24–26]. One large study reported 17 cases of LGGs that were successfully treated with a combination of neoadjuvant temozolomide and surgical removal after tumor volume reduction from among 170 recipients of upfront temozolomide [25, 26]. This study, however, reported selected cases in which temozolomide-induced tumor volume reduction enabled subsequent resection, and the original population in which the same strategy was attempted, is unknown. To the best of our knowledge, the present study is the first to report a prospective series of gliomas treated with a neoadjuvant strategy involving second-look resection after tumor volume decrease in cases of incomplete initial resection. Moreover, in contrast with the previous study of universally prescribed upfront temozolomide for LGGs [25, 26], upfront chemotherapy was applied only for cases with favorable molecular profiles in the present study. The median SPD change following chemotherapy in the present cohort was −35 % (−41 % in 19 evaluable codeleted tumors), with all but one case showing volume decrease of T2/FLAIR abnormality. Although previous studies suggest that LGGs responsive to chemotherapy may include those without favorable molecular profiles [26, 27], the volume changes from baseline in the present study were comparable to those of selected responsive cases in a previous study [26]. Thus, the molecular-guided approach can be used to select cases with a favorable response. Because the IDH mutation also predicts benefit from PCV chemotherapy in anaplastic oligodendroglial tumors [14] and temozolomide in LGGs [28], the analyses of IDH mutations may replace those of 1p19q codeletion and MGMT methylation for the selection of patients.

The median PFS in the present study was 57 months among the 22 grade II gliomas with either 1p19q codeletion or MGMT methylation, including the patients with focal anaplasia. Thus, upfront chemotherapy coupled with second-look resection for the cases with incomplete initial surgery might compare favorably with reported outcomes of chemotherapy only for 1p19q codeleted grade II tumors (e.g., median PFS: 42 months) [8]. Indeed, some of the present cases that received second-look resection (cases 1, 9, 12, 17, and 26) showed durable PFS for more than 5 years from the initiation of chemotherapy, despite initial biopsy or partial resection. In cases of incomplete initial resection, second-look resection following tumor volume decrease may improve survival outcome and offer an advantage of upfront chemotherapy over initial radiotherapy for chemosensitive LGGs. Moreover, durable PFS in several cases after second-look resections as well as the observation in case 7; recurrence was not demonstrated in the insular cortex where T2/FLAIR abnormality was initially shown (Fig. 3), may suggest that invasive glioma fronts can be reduced using chemotherapy.

Of the 13 second-look resections, removal of the residual 11C-methionine uptake was performed in three cases with extensive gliomas. Two of the three cases were progression free despite residual FLAIR abnormality (cases 5 and 26 in Table 1; Fig. 4), and radiotherapy was deferred, avoiding the potential for significant neurocognitive deterioration, especially in patients with large tumors. Adequate treatment (removal or radiotherapy) of areas with 11C-methionine PET uptake is associated with better outcome in gliomas [29–31]. Although further studies are warranted, the neoadjuvant strategy to remove residual methionine uptake following chemotherapy might delay tumor recurrence and/or malignant transformation, and could be among the therapeutic options for extensive gliomas.

Six of the 13 cases that received second-look resection showed subsequent recurrence, and all but one of these occurred at incomplete surgical margins on postoperative MRI, reiterating the importance of local tumor control. It should be noted that recurrence may occur even after second-look resection if the surgical margin is incomplete. Thus, second-look resection after chemotherapy should be indicated with discretion after consideration of anticipated resection volume and survival/functional benefit.

In conclusion, the present study demonstrates the utility of molecular-guided upfront chemotherapy with the intention of second-look resection for incomplete initial resections. Further studies are warranted, and optimization of the molecular criteria, delineation of the methionine uptake to be resected, and preoperative molecular diagnosis by imaging [32] would help improve the utility of the neoadjuvant strategy for gliomas. In clinical trials with a chemotherapy-alone arm, second-look resection after tumor shrinkage may be considered for incomplete initial resection.