Introduction

The tumor resection rate, and the age and performance status of the patient, are major prognostic factors in determining the life expectancy of patients with diffuse glioma1. However, because of the invasive nature of gliomas, the boundary between tumor tissue and the surrounding normal brain tissue is often unclear, and a very high level of surgical experience is required to determine the appropriate interface to achieve maximal safe resection and functional preservation. Although in situ visualization of tumor tissue with 5-aminolevulinic acid fluorescence is widely used in surgery of high-grade gliomas, the degree of fluorescence differs considerably between cases, and its utility can decrease in some recurrent cases2,3. The development of a surgical device to more objectively differentiate tumor tissue from normal parenchyma may aid surgeons during neurosurgery.

Haptics is a technology that measures and transmits a sensation caused by mechanical contact with an object, and the haptic sensation can be quantified, amplified, and reproduced4,5,6,7,8,9. Previous studies have reported that cancer stiffness was useful in diagnosis and grade prediction in colorectal cancer, and correlated with pathological findings and molecular expression in cancer tissue10,11. Cancer stiffness may also be useful for prediction of complications after pancreatic cancer surgery and the occurrence of lymphedema after breast cancer surgery12,13,14. Thus, cancer stiffness may be a useful clinical biomarker in the management of cancer patients.

We previously developed a prototype of a leader–follower microsurgical forceps equipped with haptic technology (hereafter termed forceps), and examined whether it could distinguish the stiffness of two kinds of Tofu (soybean curd)—silk Tofu (soft type) and firm Tofu—as a model of brain tissue distinction15. We found that the forceps could reproducibly differentiate the two types of Tofu for each of the Tofu makers. Herein, as a prestage for clinical introduction, we examined the clinical utility of our haptics forceps to distinguish between brain tumors and normal brain tissues, as well as between different tumor pathologies, by assessing the stiffness of glioblastoma, malignant meningioma, and normal brain tissues in mice.

Results

The median tumor stiffness of the glioblastoma cell lines was SF126-FmC = 210.30 N/m, U87-FmC = 218.94 N/m, and U251-FmC = 215.71 N/m. The median tumor stiffness of the malignant meningioma cell line was IOMM-Lee-FmC = 269.53 N/m and HKBMM-FmC = 192.26 N/m. By contrast, the median stiffness of normal brain tissue was 155.84 N/m (Fig. 1, upper panel; Supplementary Table S1). Both glioblastoma (SF126-FmC, U87-FmC, U251-FmC; n = 5 per cell line [n = 15 total]; p < 0.001) and malignant meningioma (IOMM-Lee-FmC, HKBMM-FmC; n = 5 per cell line [n = 10 total]; p < 0.001) were significantly stiffer than normal brain tissue (n = 25; Fig. 1, lower panel).

Fig. 1
figure 1

Analysis of tumor and brain stiffness. The Kruskal–Wallis test was used to compare stiffness between the normal brain (blue box, n = 25) and the SF126-firefly luciferase-mCherry (FmC) (red box, n = 5), U87-FmC (red box, n = 5), U251-FmC (red box, n = 5), IOMM-Lee-FmC (green box, n = 5), and HKBMM-FmC (green box, n = 5) tumors if there were statistically significant overall differences (upper panel). The Kruskal–Wallis test was performed to compare the stiffness of the normal brain (n = 25), the three types of glioblastoma (SF126-FmC [red box, n = 5], U87-FmC [red box, n = 5], and U251-FmC [red box, n = 5]; i.e. n = 15 in total), and the two types of malignant meningioma (IOMM-Lee-FmC [green box, n = 5] and HKBMM-FmC [green box, n = 5]) if there were statistically significant differences between the normal brain, glioblastoma, and malignant meningioma tissues (lower panel).

Full size image

Next, we compared the stiffness of the left brain (brain tumor tissue) and the right brain (normal brain tissue) in each of the five mice. Each type of brain tumor was also significantly stiffer than normal brain tissue (SF126-FmC, p = 0.005; U87-FmC, p = 0.005; U251-FmC, p = 0.005; IOMM-Lee-FmC, p = 0.005; HKBMM-FmC, p = 0.005; Supplementary Fig. S1, upper and middle panel; Supplementary Table S2). There were no differences in stiffness between the different types of glioblastomas (SF126-FmC, U87-FmC, U251-FmC; p = 0.468). However, for the two meningiomas, the IOMM-Lee-FmC tumors were significantly stiffer than the HKBMM-FmC tumors (p = 0.032; Supplementary Fig. S1, lower panel; Supplementary Table S1). Histologically, compared with IOMM-Lee tumors, HKBMM tumors showed abundant necrotic foci (i.e., similar to glioblastoma), which may explain the differences in stiffness (Fig. 4, lower panel; Supplementary Fig. S2).

Discussion

The main finding of the present study was that our forceps equipped with haptic technology could measure the tissue stiffness and discriminate brain tumor tissue from normal brain tissue. These findings suggest that stiffness may be a useful marker for differentiating tumors. Indeed, both glioblastomas and malignant meningiomas were significantly stiffer than normal brain tissue. There were also stiffness differences between the two types of malignant meningiomas, which may reflect their pathological characteristics. Thus, tumor stiffness may also be useful for differentiating the histological and molecular profiles of tumors. Indeed, the HKBMM-FmC tumors showed abundant necrosis (stiffness comparable to the three glioblastomas) and were significantly softer than the IOMM-Lee meningiomas (Fig. 1). Note that there was no difference in stiffness between glioblastoma and malignant meningioma, which may relate to the small number of available brain tumors in our study. In addition, it has been suggested that the stiffness of the tumor varies depending on the heterogeneity of the tumor, the boundary with the normal brain, and the area of edema16,17. However, this cannot be analyzed in the mouse model used in the present study, so improvements to the device and the establishment of a new experimental system such as a human holding the device are required.

Hardness of certain substance is commonly evaluated using parameters such as Young’s modulus. However, standard method for measurement of tissue stiffness has not been established yet. In the literature, some investigators measured the stiffness of colorectal cancer using the contact impedance method13. However, the device is too large to measure brain tissue. We sought to develop microsurgical forceps equipped with haptics technology specialized for brain surgery that enables measurement of tissue stiffness. In the present study, the relative stiffness was measured by the forceps being fixed on the base. The forceps was automatically closed by computer program without examiner operation, maintaining the same opening width and speed of closure; the contact area between the forceps and tissue was thus assumed to be consistent in all measurements, allowing relative comparison between tissues. The measurement may not work in the same way if the forceps was held by the examiner, and future improvement of the system is warranted.

During brain tumor surgery, surgeons determine the interface between brain tumors and the surrounding parenchyma by stiffness, color, and vasculature, and experience is required for appropriate differentiation between tumor tissue and normal tissue. As it has been reported that the force profile of surgical procedures varies depending on the years of experience of neurosurgeons, Standardization of surgery is necessary for safe surgery18. A surgical navigation system and intraoperative MRI are commonly utilized in brain tumor surgery19,20,21. Nevertheless, the actual in situ dissection is conducted on the basis of the surgeon’s sensation while referring to the information provided by the non-in situ devices. Fluorescence-based diagnosis using 5-aminolevulinic acid provides information for in situ dissection22. However, its utility is mostly limited to glioblastoma patients, and the degree of fluorescence significantly differs between cases. Although intraoperative rapid pathological diagnosis based on frozen tissue is important, it can only be performed intermittently and takes time for analysis. Given that our forceps equipped with haptic technology could distinguish between tumors and normal brain tissue, it may aid neurosurgeons in determining the dissection interface when used together with the aforementioned techniques. Moreover, the forceps may aid the intraoperative pathological diagnosis of tumors with machine-learning/artificial intelligence (AI) technology. Because the force–tactile sensation could be amplified, reproduced, and transmitted, the accuracy for measuring tumor stiffness could be improved. Furthermore, the forceps may be useful for surgical education. Haptic feedback has the potential to be applied in remote robotic surgery as it allows the surgeon to simultaneously perceive and quantify subtle differences in tissue stiffness. Moreover, neurosurgical devices equipped with haptic technology have shown potential for application in surgery such as hemangioblastoma and arteriovenous malformations (AVM), suggesting future applications in neurosurgery23,24. Finally, the forceps and technology could be applied to the development of surgical robots with remote controls.

In conclusion, the present study showed that our microsurgical forceps equipped with haptic technology could distinguish between tumors and normal brain tissue by stiffness. This technology may help to improve safety and the resection rate in brain tumor surgery and contribute to improvement and uniformity of surgical results regardless of the surgeon’s experience.

Materials and methods

Cell lines

Human glioblastoma cell lines (SF126, U87, and U251) and human malignant meningioma cell lines (IOMM-Lee and HKBMM) were used. SF126, U87, and U251 were provided by Dr. Sampetrean’s group. IOMM-Lee was provided by Dr. Kitamura. HKBMM was manufactured at Cosmo Bio Co. Ltd., Tokyo, Japan. The cell lines were treated with 10% (vol/vol) fetal bovine serum (GIBCO Thermo Fisher, Tokyo, Japan) and 1% (vol/vol) penicillin/streptomycin (Antibiotic–Antimycotic, 100×; Fujifilm Wako Pure Chemicals Co. Ltd., Osaka, Japan) in Dulbecco’s modified Eagle’s medium (GIBCO Thermo Fisher, Tokyo, Japan). Each cell line was modified to express firefly luciferase-mCherry (FmC) protein, and were named SF126-FmC, U87-FmC, U251-FmC, IOMM-Lee-FmC, and HKBMM-FmC25.

Glioblastoma and malignant meningioma mouse models

This study was approved by the ethics committee of Keio University School of Medicine (#20200325) and was conducted in accordance with Animal Research: Reporting In Vivo Experiments (ARRIVE) guidelines. In vivo animal experiments were approved by the Keio University Institutional Animal Care and Use Committee (#18053-2) and were conducted in accordance with the institutional guidelines on animal experimentation at Keio University.

Five female ALB/cSlc-nu/nu nude female mice 6 weeks old (Sankyo labo service cooperation Inc., Tokyo, Japan) were injected intraperitoneally with a triad of anesthetics (medetomidine 0.3 mg/kg + midazolam 4 mg/kg + butorphanol 5 mg/kg) and secured to a stereotactic brain surgery device. The scalp was incised and a hole was opened in the skull covering the right cerebral hemisphere (2-mm lateral and 1-mm anterior to bregma). The SF126-FmC, U87-FmC, U251-FmC, IOMM-Lee-FmC, and HKBMM-FmC cells (5.0 × 105 cells/3 μL phosphate buffered saline [PBS] per cell line) were implanted into the right frontal cortex at a depth of 0.5 mm via the skull hole using a Hamilton micro syringe (Fig. 2). The wound was closed with 6–0 nylon thread, and the mice were kept awake in the supine position with heating after the procedure26. A timer was used to adjust the light exposure at the same time as the external environment in a clean animal room using a High Efficiency Particulate Air (HEPA) filter (Cambridge filter cooperation Ltd., Tokyo, Japan). For hygiene reasons, food, water, and cages were changed two times a week.

Fig. 2
figure 2

Generation of mouse models of brain tumors and cell transplantation. Brain tumor cells (5.0 × 105 cells/3 μL phosphate buffered saline [PBS]) were implanted intracranially into anesthetized ALB/cSlc-nu/nu nude mice at a rate of 1 μL/min. Luciferin was administered into the abdominal cavity before the tumor growth process was measured by an in vivo imaging system (IVIS).

Full size image

In vivo imaging

After transplantation of glioblastoma and malignant meningioma cells, in vivo brain tumor growth was observed once a week by using the IVIS system (PerkinElmer IVIS® Lumina LT In Vivo Imaging System; Promega Cooperation, Tokyo, Japan). A total of 150 µL luciferin (Fujifilm Wako Pure Chemicals Co. Ltd.) was administered intraperitoneally to mice, and they were anesthetized using isoflurane (Pfizer Japan Inc. Tokyo, Japan) inhalation anesthesia to induce sleep. After seven minutes, imaging was performed using IVIS. After imaging, mice were confirmed to regain consciousness to prevent accidental death (Figs. 2 and 3).

Fig. 3
figure 3

Microsurgical forceps equipped with haptic technology and measurement of tissue stiffness of mice. Upper panel: Forceps equipped with our haptics technology. Lower right panel: After euthanasia, brain tissues were removed under optical and fluorescence microscopy. Measurements were also taken by inserting the forceps as shown in the figure.

Full size image

Microsurgical forceps equipped with haptic technology

Stiffness of the brain tumor tissue and normal brain tissue were measured using a leader–follower forceps prototype (hereafter termed forceps; Fig. 3). The forceps was equipped with two voice coil motors (AVM19-5; Akribis Systems Co., Ltd., Tokyo, Japan), which was controlled by a bilateral control system4,27. Via bilateral control method, the haptic information can be transmitted and be recorded. Then, the relative stiffness of grasping object through the haptic forceps can be calculated from the force information estimated by disturbance observer and the position information detected by the optical encoder.

Measurements

Mice were sacrificed when neurological symptoms such as anorexia or gait disorder became apparent. Euthanasia was performed by administering an anesthetic (intraperitoneally) to induce sleep, followed by administering a lethal dose of anesthetic. The brains were rapidly removed, fixed overnight in 4% paraformaldehyde + PBS (Fujifilm Wako Pure Chemicals Co. Ltd.), washed with 1 × PBS (pH 7.4), and stored at 4 °C until measurement. Paraformaldehyde fixed brain tissue was measured by the forceps. The location of the tumor was confirmed using optical and fluorescence microscopy, and the forceps (tip width = 4 mm) was inserted into the tumor at a depth of 5 mm (Fig. 3). The tumors were located in the right hemisphere, and were assessed for stiffness. The stiffness of the normal brain tissue was measured in the left hemisphere as a control. In the experiments, the silver tip of the forceps is used as a marker for insertion depth (5 mm) and the forceps was narrowed automatically with the constant speed by computed program from the same starting width (4 mm); not by examiner holding and operating. The body of the forceps is secured to the base with a clamp; the contact area of the forceps to tissue was thus assumed to be as consistent as possible (Fig. 3). The stiffness is measured between the two points where the forceps start closing and where they finish closing. The reaction force (N) as the force response, and stiffness (N/m) as the position response, were measured (Figs. 3 and 4, upper panel).

Fig. 4
figure 4

Stiffness measurement and pathological imaging. Upper panel: Relationship between force versus time and force versus distance. Two points were used to calculate the reaction force and stiffness from the slope. Lower panel: Hematoxylin and eosin (HE) staining of specimens to examine for tumor formation after tumor cell transplantation. Magnification, 200×.

Full size image

The primary outcome is to determine whether forceps have the ability to distinguish between the stiffness of brain tumor and normal brain tissue.

Tissue staining

After the forceps measurement, hematoxylin and eosin staining was performed for histological confirmation (Fig. 4, lower panel; Supplementary Fig. S2). The secondary outcome is to ascertain whether a correlation exists between the stiffness and pathological characteristics when there is a significant difference in stiffness among brain tumors with the same tumor grade.

Statistics

Stiffness (N/m) was analyzed as a parameter of the hardness of brain tumor tissue and normal brain tissue. Differences in the stiffness between normal brain tissue (n = 25), glioblastoma tissue (SF-126-FmC, U87-FmC, and U251-FmC; n = 5 for each [total of n = 15]), and malignant meningioma tissue (IOMM-Lee-FmC and HKBMM-FmC; n = 5 for each [total of n = 10]) were evaluated by the Kruskal–Wallis test with independent samples. Next, we compared the stiffness of the three types of glioblastoma (n = 15) versus the two types of malignant meningioma tissues (n = 10), the stiffness of the normal brain (n = 25) versus the glioblastoma (n = 15), and the stiffness of the normal brain (n = 25) versus the malignant meningioma (n = 10) using the Mann–Whitney U test. Finally, the stiffness of the tumors was compared with normal brain tissue for each brain tumor type using the Wilcoxon signed-rank test, while differences in stiffness between the three types of glioblastoma and the two types of malignant meningioma were evaluated by the Kruskal–Wallis test. All statistical analyses were performed using statistical software (IBM SPSS statistics 25, Chicago, IL, USA), while graphs were created with Graph Pad Prism (GraphPad, La Jolla, CA, USA). A p value < 0.05 was considered statistically significant. This study was not randomized and mice with sufficiently grown and measurable tumors were used. Mice with tumors that were too small were excluded because they could not be measured using forceps.