Clinical characteristics of the analyzed cases

The clinical characteristics of the 39 cases analyzed, including sex and age, are described in Table 1. In all the cases, pathologic diagnosis or integrative diagnosis with pathologic and molecular analysis of the tumor specimens obtained by initial biopsy or tumor removal was glioblastoma according to the World Health Organization (WHO) classification revised 4th edition [11] or glioblastoma, IDH-wildtype according to the WHO classification 5th edition [1]. A total of 20 and 19 cases treated at the Nagoya University Hospital and Shizuoka Cancer Center, respectively, were included. The pre-operative symptoms were hemiparesis (n = 12), seizure (n = 10), sensory aphasia (n = 9), headache (n = 5), and motor aphasia (n = 4). The primary tumor locations were the frontal (n = 17), temporal (n = 10), and parietal (n = 4) lobes. The median tumor volume was 27.32 mL (11.77–44.55 mL). Initial surgery was performed as follows: biopsy (n = 4), gross total resection (GTR; n = 22), subtotal resection (STR; n = 9), and partial resection (PR; n = 4).

Table 1 Overview of patient and tumor characteristicUsage rate and length of TTFields treatment

Among the 39 cases, TTFields treatment was discontinued in 18 cases as requested by the patients (discontinued group), while TTFields treatment continued in 21 cases until progression or completion, or have been continuing TTFields treatment (continued group). Patient-requested discontinuations included only the patient who did not attempt to complete the recommended TTFields treatment and discontinued at the patient's request even without tumor recurrence. The clinical characteristics of these cases are described in Table 2. Patients in the continued group underwent TTFields treatment for longer periods than those in the discontinued group (median 385 days (241–487 days) and 42 days (23–152 days), respectively; p < 0.001). Also, patients in the continued group showed significantly higher usage rates than those in the discontinued group (median 84% (65–90%) and 49% (0–79%), respectively; p < 0.001). Comparing various clinical findings between these subgroups, older cases were more frequently found in the discontinued group (p = 0.019). Also, pre-operative mild or severe sensory aphasia symptoms were observed more frequently in the discontinued group than in the continued group (eight cases in the discontinued group and no case in the continued group, p < 0.001). Temporal lobe tumor in the dominant hemisphere was more frequently observed in the discontinued group than in the continued group, although this observation was not statistically significant. (p = 0.153). The extent of resection and KPS score was not significantly different between these groups. These findings suggested that sensory aphasia symptom is correlated with low usage rate and discontinuation as requested by the patients together with older age. Median PFS by the Kaplan–Meier method was 336 days in the continued group and 290 days in the discontinued group (p = 0.653), and median OS was 632 days and 593 days, respectively (p = 0.552). There was no statistically significant difference in PFS/OS between the continued and discontinued groups, possibly due to the short observation period and small number of cases. However, identification of the factors leading to discontinuation and low usage rates is important to help derive maximum benefit from the efficacy of TTFields demonstrated in the previous reports.

Table 2 Usage condition of TTFieldsTumor recurrence in the continued group

Among 21 cases in the continued group, recurrence during TTFields treatment was observed in 15 cases. Of these 15 cases, 10 cases exhibited local recurrence around the primary tumor and one of the 10 cases revealed additional leptomeningeal dissemination. By contrast, in five cases, patients exhibited recurrence in the distant parenchyma from the primary lesion. Compared with data obtained from a previous study which reported the recurrence pattern of glioblastoma treated without TTFields treatment, distant parenchymal recurrence was more frequently observed in these 15 cases (p < 0.01) [12, 13]. Among the five cases who exhibited recurrence in the distant parenchymal from the primary lesion, one, two, and two cases exhibited recurrence in another lobe of the ipsilateral hemisphere, in the contralateral hemisphere, and the cerebellum, respectively. In these five cases, higher usage rates and long-term use were recorded compared to the 10 cases (median 88% (73–89%) vs 65% (51–75%) and median 525 days (252–1361 days) vs 238 days (108–329 days), p = 0.019 and p = 0.040, respectively). Also, more favorable PFS were observed in these cases compared to other cases (p = 0.024; Fig. 1). Median PFS by the Kaplan–Meier method was 525 days in the group of distant parenchymal recurrence and 196 days in the group of non-distant parenchymal recurrence, and median OS was 679 days and 548 days, respectively (p = 0.141; supplementary Fig. 1). There was a trend toward longer OS in the distant parenchymal group, although there was no statistically significant difference in OS due to the short observation period and small number of cases. These findings suggested that atypical recurrence such as those in distant parenchyma may be observed in cases where TTFields treatment is used for long periods.

Fig. 1

Kaplan–Meier curve indicating PFS. Kaplan–Meier curve indicating PFS of patients with distant parenchymal recurrence (n = 5; blue line) and patients without distant parenchymal recurrence (n = 10; orange line; p = 0.024)

Two characteristic cases exhibiting recurrence in distant parenchyma after long-term TTFields treatmentCase 1

In a 71-year-old male, an MRI revealed a small gadolinium-enhanced lesion in the left temporal lobe (Fig. 2A). Biopsy analysis revealed glioblastoma, IDH-wildtype, based on both histologic and molecular evaluation of the tumor tissue. Immediately after the biopsy, he underwent radiation therapy (60 Gy/30 fr) concomitant with TMZ. After that, he started maintenance therapy using TTFields treatment with TMZ (Fig. 2B). He could maintain stable disease (SD) status continuing TTFields treatment with TMZ. However, after 17 months, an MRI revealed a recurrence of a gadolinium-enhanced lesion in the right cerebellum hemisphere, while the primary lesion maintained SD status (Fig. 2C). BEV treatment was introduced; however, 25 months after the initial biopsy, the patient died owing to progression of the tumor (Table 3).

Fig. 2

MRI and Methionine-PET images of Case 1. (A) Pre-operative MRI contrast-enhanced T1-weighted image (left) and Methionine-PET (right). Yellow arrow indicates tumor lesion. (B) MRI contrast-enhanced T1-weighted image after radiation therapy concomitant with TMZ. Yellow arrow indicates tumor lesion. (C) MRI contrast-enhanced T1-weighted image (left) and Methionine-PET (right) of the primary lesion (upper) and cerebellum (lower) on recurrence. Yellow arrows and arrow heads indicate primary lesions and recurrent lesions, respectively

Table 3 Recurrence during TTFields treatmentCase 2

A 42-year-old male developed left hemiparesis and partial seizure. MRI revealed a gadolinium-enhanced mass lesion in the right frontal lobe. GTR was achieved and the integrative diagnosis was glioblastoma, IDH-wildtype (Fig. 3A, B). Immediately, he underwent radiation therapy (60 Gy/30 fr) concomitant with TMZ. After that, he started maintenance therapy using TTFields treatment with TMZ. For 67 months, he could maintain complete response (CR) status by continuing TTFields treatment with TMZ. MRI and Methionine-PET revealed recurrent tumors around the primary tumor lesion together with those in the right temporal lobe (Fig. 3C). Tumor removal for both lesions was performed. Pathologic findings of recurrent tumors around the primary lesion and temporal lobe were consistent with those of grade III glioma with 30% of Ki67 index and grade II glioma with 5–10% of Ki67 index according to the WHO classification revised 4th edition, respectively (Fig. 3D). Both tumors revealed TERT promoter mutation, CDKN2A homozygous deletion and EGFR amplification without IDH mutation. An electric fields map provided by Novocure revealed that the local minimum power density (LMiPD) of the TTFields was lower in the temporal lesion compared to those in the primary lesion (Fig. 4). Because temporal lesions were outside of area treated with prior radiation therapy, after tumor removal, we performed radiation therapy for temporal tumor as recurrent glioblastoma, IDH-wildtype. These cases suggested that during long-term use of TTFields therapy, not only local recurrence around primary lesion but distant recurrence can occur outside the area targeted by TTFields, such as in the inferior tentorial lesion and in areas with lower TTField exposure.

Fig. 3

MRI, Methionine-PET, and Hematoxylin and Eosin (HE) images of Case 2. (A) Pre-operative (left) and post-operative (right) MRI contrast-enhanced T1-weighted image. Yellow arrows indicate tumor lesion. (B) HE image of primary tumor. A scale bar indicates 100 um. (C) MRI contrast-enhanced T1-weighted image (left), FLAIR images (middle), and Methionine-PET (right) of the recurrent lesion. Axial image (upper and middle) and coronal image (lower). Yellow arrows and arrow heads indicate recurrent tumors around primary lesions and on distant parenchyma, respectively. (D) HE image (left) and IHC using anti-Ki67 antibody (right) of recurrent tumor around the primary lesion (upper) and on distal parenchyma (lower). Scale bars indicate 100 um

Fig. 4

Electric fields map for recurrent lesions of Case 2. (A) Axial (left) and sagittal (right) images of electric fields for a recurrent tumor on the primary lesion. The purple, red or green line indicates gross tumor volume (GTV; 1.8 LMiPD), clinical target volume (CTV; 1.5 LMiPD) and peritumoral boundary zone (PBZ; 1.5 LMiPD), respectively. (B) Axial (left, upper), sagittal (right, upper), and coronal (left, lower) image of electric fields for recurrent tumor on distant parenchyma. The purple line indicates GTV (1.1 LMiPD) of a recurrent lesion