Hyperintense signal on diffusion-weighted imaging for monitoring the acute response and local recurrence after photodynamic therapy in malignant gliomas

Photodynamic therapy (PDT) subsequent to surgical tumor removal is a novel localized treatment for malignant glioma that provides effective local control. The acute response of malignant glioma to PDT can be detected as linear transient hyperintense signal on diffusion-weighted imaging (DWI) and a decline in apparent diffusion coefficient values without symptoms. However, their long-term clinical significance has not yet been examined. The aim of this study was to clarify the link between hyperintense signal on DWI as an acute response and recurrence after PDT in malignant glioma. Thirty patients (16 men; median age, 60.5 years) underwent PDT for malignant glioma at our institution between 2017 and 2020. We analyzed the signal changes on DWI after PDT and the relationship between these findings and the recurrence pattern. All patients showed linear hyperintense signal on DWI at the surface of the resected cavity from day 1 after PDT. These changes disappeared in about 30 days without any neurological deterioration. During a mean post-PDT follow-up of 14.3 months, 19 patients (63%) exhibited recurrence: 10 local, 1 distant, and 8 disseminated. All of the local recurrences arose from areas that did not show hyperintense signal on DWI obtained on day 1 after PDT. The local recurrence in malignant glioma after PDT occurs in an area without hyperintense signal on DWI as an acute response to PDT. This characteristic finding could aid in the monitoring of local recurrence after PDT.


Introduction
Survival is associated with the extent of resection in patients with glioblastoma, the most aggressive primary malignant brain tumor [1][2][3][4]. However, optimal surgical treatment that balances maximal resection and safe resection is still challenging in glioblastoma because of its incredibly invasive nature [5]. Standard therapy includes maximal possible surgical resection followed by radiotherapy and concomitant temozolomide-based chemotherapy and often leads to local recurrence [6][7][8][9][10]. Photodynamic therapy (PDT) subsequent to surgical tumor removal is a novel localized treatment for malignant glioma and a possible alternative to standard therapy. PDT is a light-activated treatment modality that provides effective local control [11][12][13]. In PDT, a photosensitizer that is intravenously injected before surgery and selectively accumulates in the tumor is transformed from its ground state into an electronically excited state by semiconductor laser of a specific wavelength; this transfers the energy directly to triplet oxygen to form singlet oxygen (a type II reaction), which exerts an antitumor effect specifically in tumor cells while preserving the surrounding normal brain cells [14][15][16]. However, even PDT cannot completely suppress local recurrence [11,12].
The treatment outcome of glioblastoma at recurrence is devastating with all known types of therapy [17] but second surgery to treat recurrence is considered a valid option [18][19][20][21][22]. Reoperation in selected patients with a favorable Karnofsky Performance Status (KPS) score at the time of recurrence significantly improves overall survival from recurrence without a significant decrease in functional status [21,22]. To enable second surgery as a treatment option for recurrence, the early detection of the resectable recurrence before the patient's KPS score decreases is crucial.
We previously reported that the acute response of malignant glioma to PDT can be detected as transient linear hyperintense signal on diffusion-weighted imaging (DWI) and a decline in apparent diffusion coefficient (ADC) values without symptoms [23]. However, the long-term clinical significance of these imaging findings was not examined. Outcomes would be improved if recurrence could be predicted earlier from this characteristic PDT-induced change on magnetic resonance (MR) images. We hypothesized that the characteristic hyperintense signal on DWI could help not only to confirm the PDT-irradiated area, but also to monitor the local recurrence site after PDT in malignant glioma. The aim of this study was to clarify the link between hyperintense signal on DWI as an acute response and recurrence after PDT in malignant glioma.

Study design
This study was approved by the institutional review board (protocol number B190100) and conducted according to institutional and national ethical guidelines and in accordance with the Helsinki Declaration.
At our institution, PDT has been combined with conventional surgery since August 2017. The target patients for PDT are those with confirmed World Health Organization (WHO) grade III or IV glioma by rapid intraoperative pathological diagnosis and those who underwent more than 90% (gross total or subtotal) resection of contrast-enhancing lesions or intentional partial resection due to involvement of eloquent areas. Thirty consecutive patients underwent PDT for malignant glioma at our institution between August 2017 and December 2020. We analyzed the signal changes on DWI after PDT and the link between these findings and the recurrence pattern.

PDT procedure
1. Preoperative: Patients preoperatively expected to have a higher-grade glioma were scheduled for PDT. These patients were administered a single intravenous injection of talaporfin sodium (Laserphyrin; Meiji Seika Pharma Co., Ltd., Tokyo, Japan) at a dose of 40 mg/m 2 22-26 h before surgery. 2. Intraoperative: The histological malignancy (WHO grade III or higher) was confirmed by rapid intraoperative pathological diagnosis. After maximum resection of the contrast-enhancing lesion, complete hemostasis was ensured before PDT irradiation. Next, the entire resection cavity was irradiated with a 664-nm semiconductor laser (Meiji Seika Pharma Co., Ltd.) (diameter, 1.5 cm; radiation power density, 150 mW/cm 2 ; radiation energy density, 27 J/cm 2 ). Each irradiation was performed for 3 min per spot without overlap. The large blood vessels were protected by aluminum foil to avoid their direct irradiation ( Supplementary Fig. 1). 3. Postoperative: For the prevention of photosensitive dermatosis, post-therapeutic light protection (< 500 lx) was performed for 10-14 days.
As post-PDT therapy, radiotherapy plus concomitant and adjuvant temozolomide (Stupp regimen) were typically started as soon as the final histological and molecular/ genetic integrated diagnosis was confirmed.

Imaging analysis
Post-PDT imaging follow-up was performed using a 3.0-T MR imaging scanner (Achieva; Philips Medical Systems, Eindhoven, The Netherlands) on days 1, 7, 14, 30, and 60 and every 3 months after that. Extra MR imaging was performed at the time of neurological deterioration. matrix, 304 × 304; slice thickness, 0.8 mm) before and after injection of intravenous gadolinium contrast agent (0.2 ml/ kg; Magnescope, Guerbet, Paris, France). In this study, recurrence was defined as follows: local recurrence, growth of a contrast-enhancing lesion at the primary tumor bed; distant recurrence, growth of a contrast-enhancing lesion that appeared separately from the primary tumor bed; dissemination, arachnoid or ependymal enhancement away from the primary tumor site or spinal metastasis. To distinguish between recurrence and pseudoprogression, we also defined pseudoprogression according to the Response Assessment in Neuro-Oncology criteria as follow: new or enlarging contrast-enhancing lesion occurring within 12 weeks after the completion of radiotherapy in the absence of true progressive disease. Recurrence was assessed by two independent reviewers (KT and TS). The link between the signal changes on DWI obtained on day 1 after PDT and the recurrence site was analyzed by three independent reviewers (YF, HN, and TS).

Statistical analysis
The characteristics of recurrent and non-recurrent patients were compared using Fisher's exact test and the Mann-Whitney U test. The Kruskal-Wallis test with the Steel-Dwass post hoc test was used to compare the characteristics of patients based on recurrence patterns. All statistical analyses were performed with EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R software (http:// www.r-proje ct. org/) [24]. A two-sided p-value < 0.05 was considered statistically significant.

Recurrence after PDT
Of the 30 patients, 19 (63%) exhibited recurrence-local recurrence in 10 (33%), distant recurrence in 1 (3%), and dissemination in 8 (27%)-during a median follow-up period of 14.3 (range, 3.4-45.9) months. The tumor characteristics based on the recurrence pattern after PDT are summarized in Table 2. There was no significant difference in the baseline characteristics between recurrent and non-recurrent patients, except for the higher MIB-1 index in the recurrent patients (Supplementary Table 1). There was also no significant difference in the patient characteristics between the recurrence patterns (Supplementary Table 2).

Post-PDT MR imaging
In all patients, linear hyperintense signals on DWI were detected at the surface of the resected cavity from day 1 after PDT (Figs. 1, 2, 3 and 4). All hyperintense signals on DWI disappeared in about 30 days without any neurological deterioration ( Supplementary Fig. 2).

Case 1
A 49-year-old right-handed woman presented with episodes of convulsive seizures. Preoperative MR imaging revealed a contrast-enhancing mass lesion with a necrotic area in the left frontal lobe (tumor volume, 17.8 cm 3 ) (Fig. 1a, upper row). Awake surgery was performed because the tumor was close to Broca's area. Gross total resection of the contrastenhancing lesion was achieved without aphasia. After the histological malignancy was confirmed in rapid intraoperative pathological diagnosis, PDT was performed (irradiation spots, 11). Postoperative MR images obtained on day 1 after PDT showed complete resection and hyperintense signal on DWI adjacent to the resection cavity wall without symptoms (Fig. 1a, middle row). The final histological and molecular/genetic integrated diagnosis was WHO grade IV glioblastoma, IDH wild-type. She underwent radiotherapy, concomitant and adjuvant temozolomide, and tumor-treating fields therapy. However, she was found to be allergic to temozolomide in one cycle of adjuvant temozolomide and was switched to the PAV regimen. After three cycles of the PAV regimen, her chemotherapy was switched to bevacizumab alone due to paralytic ileus, an adverse effect of vincristine. Eventually, she exhibited local recurrence 22.5 months after surgery (Fig. 1a, lower row). The recurrence site was the area that did not show a hyperintense signal on DWI obtained on day 1 after PDT (Fig. 1b). She continued bevacizumab and tumor-treating fields therapy for the recurrence and had a good KPS score (90) 5.5 months after recurrence at final follow-up.

Case 2
A 67-year-old right-handed woman presented with headache and left homonymous hemianopia. Preoperative MR images revealed a ring-enhancing mass lesion in the right temporal lobe (tumor volume, 60.6 cm 3 ) (Fig. 2a, upper row). After the histological malignancy was confirmed in rapid intraoperative pathological diagnosis, the contrast-enhancing lesion was completely resected and PDT was performed (irradiation spots, 29). Postoperative MR images obtained on day 1 after PDT showed complete resection and hyperintense signal on DWI adjacent to the resection cavity wall without symptoms (Fig. 2a, middle row). The final histological and molecular/ genetic integrated diagnosis was WHO grade IV glioblastoma, IDH wild-type. She underwent radiotherapy, concomitant and adjuvant temozolomide, and tumor-treating fields therapy. Eventually, she exhibited local recurrence 7.7 months after surgery (Fig. 2a, lower row). The recurrence site was the area that did not show a hyperintense signal on DWI obtained on day 1 after PDT (Fig. 2b). She continued adjuvant temozolomide and tumor-treating fields therapy for the recurrence and had a good KPS score (70) 15.0 months after recurrence at final follow-up.

Discussion
Recurrence is a hallmark of malignant glioma, regardless of the therapy applied [6][7][8][9][10]17]. Although PDT is a novel localized treatment for malignant glioma that can selectively kill tumor cells [14][15][16], it cannot completely prevent local recurrence [11,12]. The prognosis of malignant glioma is unfortunately poor at recurrence, but its early detection can vastly improve the clinical course  and patients' management, such as reoperation, which might boost prognosis [18][19][20][21][22]. In this study, we found that a hyperintense signal on DWI as the post-PDT acute response helped to predict the local recurrence site. The PDT-irradiated area asymptomatically showed transient linear hyperintense signal on DWI. All local recurrences arose from the areas without this hyperintense signal on DWI that did not show the acute response on day 1 after PDT. PDT exploits the tumor-selective accumulation of a photosensitizer and a photochemical reaction upon semiconductor laser irradiation. In this study, talaporfin sodium, a second-generation photosensitizer, was used for PDT without the combination of other photosensitizers including 5-aminolevulinic acid. Singlet oxygen is generated and exerts an antitumor effect in the penetration depth of the semiconductor laser (5-7 mm) [11,14]. The singlet oxygen exerts this antitumor effect without bystander effects because of its short migration distance (0.02-1 μm) and short lifetime (0.04-4 µs) [14,25,26]. Histopathologically, the effect of PDT is underpinned by direct tumor cell killing, including both apoptosis and necrosis [13,27], tumorassociated vascular damage [28,29], and activation of the immune response against tumor cells [30][31][32]. Because the PDT-induced cell damage and microcirculatory impairment leads to restricted diffusivity of water molecules, the PDT-irradiated area shows hyperintense signal on DWI and a decline in ADC values as the acute response [33]. All patients in this study exhibited linear hyperintense signal of about 5 mm in size on DWI at the surface of the resected cavity from day 1 after PDT, as we previously reported [23]. These characteristic changes asymptomatically disappeared in about 30 days.
About 80% of glioblastomas treated with the Stupp regimen develop local recurrence within 2 cm of the tumor margin [6][7][8][9][10]17]. The extent of the glioblastoma resection is undoubtedly an independent prognostic factor [1][2][3][4]. However, Ellingson et al. reported that the postoperative residual contrast-enhancing tumor volume rather than extent of resection was a significant independent prognostic factor [34]. In line with this theory, PDT was added to maximal resection to kill as many tumor cells as possible around the resected cavity. PDT was a significant prognostic factor for overall survival and local progression-free survival [11,12]. Nitta (left of b). The white arrow indicates the local recurrence site (right of b). The recurrence site is the area that does not show a hyperintense signal on DWI obtained on day 1 after PDT (b). The histopathological diagnosis was WHO grade IV glioblastoma, IDH wild-type. PDT photodynamic therapy to increased relative rates of distant recurrence or dissemination [11]. In our study, local recurrence was observed in only 33% of patients, indicating excellent local control, as in Nitta et al. However, PDT cannot completely control local recurrence, and the cause of PDT failure remains unclear. As a rule, we performed PDT irradiation avoiding the large blood vessels. Therefore, PDT non-irradiated area intentionally occurred. However, except for the intentional PDT non-irradiated area, the area that did not show a hyperintense signal on DWI obtained on day 1 despite PDT irradiation would be considered as a failure spot of PDT. Several factors have been reported to contribute to PDT resistance, including overexpression of antioxidant enzyme, activation of drug efflux pumps, degeneration of photosensitizer, and increased DNA repair capacity in glioblastoma cells [35][36][37][38][39]. Our findings suggest that local recurrence after PDT occurred from the area with no acute response to PDT. This evidence supports the belief that PDT is locally effective in areas with an acute response on DWI. Clarification of the relationship between the area without an acute response to PDT and resistance to PDT would be an issue for future studies aimed at reducing PDT failure. The acute response of malignant gliomas to PDT was detected as asymptomatic transient linear hyperintense signal on DWI. Interestingly, all local recurrences after PDT occurred from areas that did not show a hyperintense signal on DWI obtained on day 1 after PDT. This characteristic finding could help to predict the local recurrence site. Careful observation of the areas without hyperintense signal on DWI after PDT might have a potential to support other MR imaging including FLAIR and contrast-enhanced T1-weighted imaging to detect progression. This would allow us to provide patients with greater opportunities for reoperation for resectable recurrence, which would undoubtedly improve their prognosis. At the same time, elucidation of the cause of PDT failure would be a key to further improve their prognosis.
Nonetheless, our study also has several limitations. First, the study was conducted at a single institution. Second, the number of patients was small. However, the safety and effectiveness of PDT in our study was almost the same as in a phase II clinical trial and a previous report [11,12]. Third, the image comparison after PDT and at recurrence had to manually consider the morphological changes in the brain. Several conformational changes in the surgical cavity made it hard to be confident that the area of local recurrence was the same area without hyperintense signal on DWI on day 1 after PDT. However, at present, there is no imaging technique to automatically adjust for the postoperative morphological changes and make a perfect one-to-one correspondence considering the time course. To make up for this limitation, we performed MR imaging evaluations more frequently than usual in this study. Postoperative MR imaging is usually performed once before and after radiotherapy. On the other hand, we performed post-PDT MR imaging followup at least three times (day 1, 7, 14) before radiotherapy, and at least one time during radiotherapy (day 30). Furthermore, based on careful following and observation of the morphological changes in the surgical cavity on all postoperative MR imaging, we analyzed the link between the DWI hyperintensity site on day1 after PDT and the recurrence site. To fundamentally resolve this limitation, it would be desirable to innovate imaging technology that could track postoperative morphological changes in the brain. Fourth, we could not find a significant link between the hyperintense signal on DWI as an acute response after PDT and distant recurrence or dissemination. The patients with distant recurrence or dissemination tended to have an uninterrupted hyperintense signal on DWI obtained on day 1 after PDT. However, the difference was not clear enough to allow their discrimination from the local recurrence cases. Fifth, this study could not fully examine the impact of bevacizumab on post-PDT MR imaging. Bevacizumab generally has a beneficial effect on pseudoprogression [40]. We had no patient with pseudoprogression in this study. Utilizing bevacizumab as first-line chemotherapy in almost one-third of our patients might contribute to this result. Finally, this study could not fully examine other risk factors for recurrence in malignant glioma after PDT. The primary tumors of all disseminated cases were localized near the ventricles. Ventricle contact has been associated with poor outcomes [41], whereas subventricular zone contact has been associated with dissemination [42]. Ventricular entry during resection has been linked to dissemination [43], although the connection is controversial [44,45]. We also could not sufficiently address other several confounding factors such as extent of resection, IDH status, and MGMT promoter methylation status, due to small number of patients. All these factors have a potential to impact recurrence after PDT. Therefore, further studies using larger groups of patients and including these factors are needed to validate the recurrence pattern after PDT. Nevertheless, it was clinically important that the local recurrence in malignant gliomas after PDT occurred in the area without the hyperintense signal on DWI as an acute response to PDT. To our knowledge, this is the first study to show the usefulness of visually clear hyperintense signal on DWI in the monitoring of local recurrence after PDT in malignant glioma.

Conclusions
This study provides new evidence that the local recurrence in malignant gliomas after PDT occurs in areas without hyperintense signal on DWI as an acute response to PDT. The characteristic hyperintense signal on DWI could help in the monitoring of not only the acute response to PDT, but also the local recurrence after PDT.