Abbreviations:
5-ALA — 5-aminolevulinic acid
CBF — cerebral blood flow
GFAP — glial fibrillary acidic protein
IDH1 — isocitrate dehydrogenase 1
IDH2 — isocitrate dehydrogenase 2
MGMT — 06-methylguanine DNA methyltransferase
WHO — World Health Organization
MRI — magnetic resonance imaging
PCR — polymerase chain reaction
Introduction
Fluorescence diagnosis has been widely used in surgery of malignant gliomas, since proliferating cells of these tumors selectively accumulate 5-aminolevolenic acid (5-ALA). This acid is transformed into protoporphyrin-IX that provides a fluorescent effect under irradiation with specific wavelength [1—3]. In recent years, high sensitivity of this method in surgery of meningiomas has been confirmed [4—6]. This is true in lesser extent for metastases, low-grade gliomas and some other tumors [7—10]. It is considered that intensity of brain tumor fluorescence depends on histological type, proliferative index, tumor cell density and some other factors [11]. MRI perfusion (ASL) data confirmed that volumetric blood flow velocity in glial tumors correlates with grade of their malignancy and intraoperative 5-ALA-induced fluorescence [12]. Moreover, it is known that ependyma and ventricular walls, choroid plexus and pia mater have a certain ability to accumulate fluorescent agent, while blood, brain vessels, intact brain tissue and dura mater do not accumulate fluorescent agent [1, 11, 13, 14]. At the same time, it is believed that infiltrative growth of malignant gliomas occurs along the fibers of white matter, ventricular ependyma and perivascular spaces and causes distant metastases and postoperative recurrences [8]. We have previously reported cortical vessel fluorescence in the bed of excised convexital meningioma (Grade I). However, biopsy and histological examination of the fluorescent vessel were associated with risk of circulatory disturbances in the cortex [6]. This observation justified more detailed study of fluorescence of not only tumor tissue, but also stromal and perifocal vessels after resection of glial tumors. Considering literature data and our own observations, we hypothesized vascular fluorescence in the structure of brain tumors can indicate their infiltration by tumor cells. It is potential pathway for dissemination and risk factor of continued tumor growth after surgery and adjuvant therapy.
The purpose of this work was to identify fluorescent vessels in the structure of brain gliomas, their histological structure, relationships with tumor and possible invasion of vascular walls.
Clinical observations and methods
A prospective cohort study enrolled patients with malignant hemispheric gliomas (Grade III—IV) who underwent surgery at the Burdenko Neurosurgery Center for the period from December 2018 to June 2019. Inclusion criteria: age≥18 years, MR-signs of glial tumor verified by postoperative biopsy, possible subtotal resection (>90% of the contrasted part), Karnofsky score ≥70, informed consent to Alasens injection. Exclusion criteria: history of increased photosensitivity or porphyria, liver disease or biochemical signs of liver dysfunction. Thus, there were 14 patients aged 20—78 years (8 men and 6 women). Temporal tumors were observed in 5 patients, frontal tumors — in 4 patients, parietal, occipital, frontotemporal, temporoparietal and parietal-occipital tumors — in one patient. Five patients had neoplasm in dominant hemisphere, 9 patients — in subdominant hemisphere. Glioblastoma de novo was diagnosed in 9 patients, continued tumor growth — in 5 patients. In one of these patients, temporal glioblastoma was diagnosed in 7 years after surgical and adjuvant treatment of anaplastic astrocytoma of the same localization. Administration of 5-ALA in brain tumor surgery was approved by the Ethics Committee of the Burdenko Neurosurgery Center. Moreover, clinical guidelines for the use of fluorescence diagnosis in brain tumor surgery were approved by the Board of the Association of Neurosurgeons of the Russian Federation and published in a peer-reviewed journal [15].
Methods
5-ALA (Alasence NIOPIC, Moscow, Russia) was administered orally 2 hours before surgery (20 mg/kg). Patients were protected from direct sunlight for 24 hours after surgery to prevent phototoxic effect. Microscope (Carl Zeiss OPMI Pentero, Germanу) with a fluorescent module (BLUE-400) was intraoperatively used. Neurophysiological monitoring of evoked motor or sensory potentials, as well as direct stimulation (Viking Select, Nicolet) were used depending on localization of tumor.
Fluorescence mode was switched on to confirm fluorescence of tumor tissue or vessels prior to resection if tumor was visualized on the brain surface or under cortex. Resection of tumor was carried out under a microscope (white light as a rule) using vacuum and/or ultrasonic suction. Fluorescence mode and ultrasound control were used for additional assessment of tumor boundaries and quality of resection. Tumor fragments were sent for histological and molecular genetic analysis. Fluorescent vessels within the tumor structure or in perifocal tissues were evaluated for their functional significance and resected for subsequent histological examination after resection of tumor. The possibility of vascular resection was determined considering anatomical localization and neurophysiological stability after temporary vascular occlusion. All patients underwent CT or MRI on the first postoperative day for analysis of post-resection cavity and exclusion of complications.
Histological specimens were fixed in formalin, embedded in paraffin and stained with hematoxylin and eosin. Moreover, immunohistochemical study with monoclonal antibodies to the glial GFAP marker was performed in 4 patients. Molecular genetic examination of specimens was carried out using real-time PCR. DNA codes were applied to detect mutations in the IDH1 and IDH2 genes and MGMT gene promoter methylation.
Results
Glioblastoma (Grade IV) with no mutation in the IDH1 and IDH2 genes was diagnosed in 10 out of 14 patients. Three of these patients had methylation of the MGMT gene promoter, this symptom was absent in 7 cases. Anaplastic (Grade III) gliomas were diagnosed in 4 patients (anaplastic astrocytoma — 2, anaplastic oligodendroglioma — 1, anaplastic pleomorphic xanthoastrocytoma – 1). Mutation in the IDH1 gene with methylation of the MGMT gene promoter was detected in 2 patients with Grade III gliomas, mutation in the BRAF gene — in 1 patient.
Intraoperative data showed that homogeneous or heterogeneous fluorescence of tumor tissue was observed in all 10 patients with glioblastoma. As usual, necrotic area did not fluoresce while the brightest fluorescence was typical for solid and infiltrative parts of tumor. Resection of tumor was followed by reduced fluorescence. A pale pink fluorescence or its absence was observed in perifocal zone. Vascular network of abnormal thin-walled non-fluorescent vessels was detected within the necrotic area in all patients.
Vessels with homogeneous or fragmentary fluorescent walls were detected in the bed of resected tumor in 4 out of 10 patients with glioblastoma (Fig. 1—3).
Histological and immunohistochemical examination
A detailed microscopic analysis of specimens with and without fluorescence of the vascular wall revealed vascular invasion in all 4 vascular samples with intraoperative fluorescence (Fig. 4).
Adventitial invasion was the most common. However, features of vascular infiltration differed not only in various clinical cases, but also in different parts of the same vessel. Several slices of the vascular wall were required to detect this phenomenon. For example, there was uneven adventitial invasion and tumor cells formed a tongue-like ingrowth fragment infiltrating tissues (Fig. 4a, b). As a result, structure of the adventitial layer was significantly disrupted. Cells lost their correct location parallel to vascular lumen, fibers had aberrant course. In addition, changes in the middle layer were also revealed. This thinned layer was characterized by dystrophic muscular elements with low density of cells and their scattered arrangement. Dystrophic changes also significantly affected the inner vascular layer that was especially noticeable in thinned and dysmorphic of endotheliocytes.
At the same time, other features of adventitial invasion were found in another fragment of the same vessel (Fig. 4c). There were two important morphological characteristics of invasion. First, adventitial spread of tumor cells was diffuse with involvement of almost the entire area of the outer vascular layer. As a result, almost all tissues of the outer layer turned out to be replaced by tumor infiltrates. There were only small islands of normal vascular cells with severe dystrophic changes. The second symptom was logically interconnected with the first one. Features of tumor growth determined extremely tight adherence of tumor to the middle vascular layer with severe dystrophy and disorganization of smooth muscle cells. However, tumor spread was observed almost along the entire circumference of the vessel unlike the previous severity of invasion.
However, some vessels were characterized by more severe destruction with malignant invasion of not only external, but also middle muscular layer. For example, small fragment with invasion of middle muscular layer is shown in Fig. 5a.
Moreover, there were reverse examples when the tumor formed a small infiltrative lesion in the outer vascular layer (Fig. 6).
Immunohistochemical study with monoclonal antibodies to the glial GFAP marker was performed in all 4 cases of vascular invasion. This method made it possible to clearly identify the areas of vascular invasion and confirm the earlier conclusions. Moreover, detailed assessment of adventitial layer was possible. It turned out that infiltration was denser and more severe in some cases compared to the data of simple histological study with hematoxylin and eosin staining (Fig. 4d).
Histological examination of the vessels without intraoperative fluorescence did not reveal their infiltration. The majority of morphological changes in these vessels were not primarily caused by tumor although they could be secondarily induced by neoplasm. Hypertrophy of the muscular layer was the most common finding. General pathological processes (atherosclerosis, hypertension) and tumor progression (Fig. 7)
There were signs of pathological modification of endothelial layer in those vessels without invasion. For example, significant uneven endothelial hypertrophy is shown in Fig. 7a.
Clinical observations
Observation No.1.
A 20-year-old patient has been diagnosed with glioblastoma of the right occipital lobe. Neurological symptoms were left-sided hemianopsia, cephalgic syndrome.
Resection of tumor was performed using ultrasound, neurophysiological monitoring and fluorescence navigation. There was bright fluorescence of the solid part of the tumor. Two vessels were found within the medial edge of the tumor after resection. There was homogenous and bright fluorescence of the larger vessel. Data of morphological examination of the fluorescent vessel are shown in Fig. 1. Cautery of distal and proximal segments of the vessels were followed by their resection for histological examination.
Postoperative course was uneventful. Improvement of visual fields was observed in 7 days after surgery (left-sided hemianopsia prior to surgery and lower-quadrant hemianopsia after surgery).
Histological and molecular genetic examinations confirmed glioblastoma (Grade IV). Mutations in the codon 132 of the IDH1 gene and in the codon 172 of the IDH2 gene, as well as methylation of the MGMT gene promoter were absent.
Observation No. 2.
A 59-year-old patient has been diagnosed with glioblastoma of the right parietal lobe. There was preoperative left-sided pyramidal symptoms. Microsurgical resection of glioblastoma of the right parietal lobe using neurophysiological monitoring, fluorescence and ultrasound navigation (Fiagon navigation system) was carried out on 16.01.2019. Fluorescent vessel was found within the bed of the tumor after resection. The vessel was sent for histological examination. Pre- and postoperative MRI data are shown in Fig. 8.
Histological and molecular genetic examinations confirmed glioblastoma (Grade IV). Mutations in the codon 132 of the IDH1 gene and in the codon 172 of the IDH2 gene, as well as methylation of the MGMT gene promoter were absent.
Observation No. 3.
A 63-year-old patient underwent resection of glioblastoma of the right temporal lobe at the place of residence on 09.08.2018 (Fig. 9a).
The patient was hospitalized to the Center of Neurosurgery on 20.01.2019 with cephalgic syndrome, smoothness of the left nasolabial fold, slurred speech, disorientation in time and space, loss of memory for current events, apathy. There were MRI signs of continued tumor growth with enlargement of contrast-enhanced area in the right frontotemporal and temporoparietal zones (Fig. 9b). Redo resection of glioblastoma of the right frontotemporal and right temporoparietal areas with neurophysiological monitoring, ultrasound and fluorescence navigation was performed. Bright fluorescence of the tumor was associated with fluorescence of the vessels inside the tumor and in the tumor bed (Fig. 3). These vessels were resected and sent for histological examination. Postoperative period was characterized by regression of cerebral symptoms without aggravation of focal symptoms (Fig. 9c).
Histological and molecular genetic examinations confirmed glioblastoma (Grade IV). R132H mutation in the IDH1 gene and mutation in the codon 172 of the IDH2 gene, as well as methylation of the MGMT gene promoter were absent.
Fragmental invasion (O) of the outer adventitial layer (A) does not reach the middle muscular layer (M) (Fig. 6).
Discussion
A prospective cohort study included patients with hemispherical gliomas who underwent surgery with fluorescence diagnosis. The feature of our sample was the absence of mutations in the codon 132 of the IDH1 gene and in codon 172 of the IDH2 gene. According to some authors, this finding can indicate the primary nature of glioblastoma [16]. However, secondary glioblastoma was a result of transformation in 7 years after resection of and adjuvant treatment of anaplastic astrocytoma (Grade III) in one case. Another feature of 7 out of 10 cases was the absence of methylation of the MGMT gene promoter as unfavorable prognostic factor [17, 18].
It should be noted that bright fluorescence of the tumor was observed in patients with glioblastoma de novo, secondary glioblastoma and in those undergoing redo surgery for continued tumor growth. These data are also confirmed in previous studies [8].
Clear fluorescence of the vessels was observed in 4 out of 10 cases after resection of fluorescent glioblastoma. However, there was no fluorescence of tumor vasculature. Obviously, there are vessels of different functional significance in the structure of tumor, perifocal zone and within the bed of excised tumor. These are intact vessels of the brain supplying normal cerebral structures near the tumor, vessels supplying normal tissue and the tumor and, finally, tumorous vessels de novo formed as a result of angioneogenesis [19—21]. Two out of 4 patients with anaplastic astrocytoma (Grade III) had mutation of the IDHI gene with MGMT gene methylation. However, tumor fluorescence was absent. Two other patients with intraoperative fluorescence of the tumor had no signs of MGMT gene methylation or mutations of the IDHI, IDH2 and BRAF genes. Thus, tumor fluorescence was observed in 12 out of 14 patients with high-grade gliomas (Grade III—IV). However, simultaneous fluorescence of the vessels was found only in 4 patients with glioblastomas (Grade IV). Bright intraoperative tumor fluorescence without vascular fluorescence was observed in 1 patient with continued growth of diffuse astrocytoma (Grade I) after previous surgery and radiotherapy.
Unlike normal vessels, tumor vasculature is characterized by high proliferative potential and unusual molecular properties [19, 22, 23]. This is especially true for glioblastoma. Tumor vessels do not have the typical functional features found in normal vessels. These vessels are more sinuous than normal cerebral vessels. Multiple vascular trunks end blindly and are not going into the underlying vascular elements due to their aberrant growth. As a result, there are hypoxic regions in the tumor. Moreover, vascular architectonics of neoplasms is characterized by high prevalence of arteriovenous shunts with fragile structure that increases the risk of hemorrhage [22].
Histological examination with haematoxylin and eosin staining revealed adventitial infiltration in fluorescent vessels and the absence of this phenomenon in those vessels without fluorescence. In addition, immunohistochemical study with monoclonal antibodies to the glial GFAP marker was performed. This protein is a marker of glial cells [24]. Therefore, its identification made it possible to clearly identify the areas of vascular wall infiltration and confirm the cause of vascular fluorescence. This method was valuable to confirm more significant and denser vascular infiltration compared with data of simple histological study with hematoxylin and eosin staining (Fig. 4d).
Two very important morphological features were revealed in one of the fragments of vascular wall (Fig. 4). The first one implied both focal and diffuse adventitial infiltration throughout the entire adventitial layer. As a result, adventitial layer was almost completely replaced by tumor infiltrates. Only residual small foci of normal vascular cells with severe dystrophy were observed. The second symptom was logically interconnected with the first one. Features of tumor growth determined the extremely tight adherence of tumor to the middle vascular layer with severe dystrophy and disorganization of smooth muscle cells. Unlike the previous level of infiltration, this type of tumor spread was observed almost along the entire circumference of the vessel (Fig. 4c).
Thus, histological and histochemical data confirmed that vascular fluorescence is determined by infiltration of the vessels. Therefore, this previously unknown phenomenon requires further analysis, since these vessels may be potentially followed by spread of tumor cells and continued tumor growth.
Conclusion
Fluorescence diagnosis with 5-ALA for analysis of tumor boundaries and perifocal tissues turned out to be a fundamentally new and unique approach in intraoperative diagnosis of vascular infiltration.
Another potential mechanism was described for the first time, i.e. tumor cells can infiltrate previously intact brain vessels. This mechanism can also contribute to progression of malignant process and dissemination of various types of glioma cells, in particular, glioma stem cells. The role of these cells has been intensively studied in recent years [21, 25—29].
This research raises the urgent problem of quality of resection of gliomas and need for resection of vessels accumulating 5-ALA, since vascular fluorescence may be a marker of malignant vascular invasion. Resection of this vessel eliminates only one of the possible mechanisms of further progression of glioma. Obviously, individualized approach considering functional significance of affected vessel is required in this case. Therefore, preoperative comprehensive assessment of tumor localization, its blood supply and relationship with surrounding vessels is essential. Modern angiographic, perfusion and radionuclide technologies (MRI, CT, PET) are valuable for these purposes. Accurate and comprehensive examination of specimens is essential. We have shown that sometimes only microinvasion of the vascular wall is observed and this finding may be confirmed by only histological examination of multiple histological slices. Intraoperative visualization of the vessels inside the tumor by using of optical equipment (microscope, endoscope) and 5-ALA fluorescence may be supplemented by ultrasound scanning, indigo carmine green (ICG) fluorescent dye. The prospects of intraoperative use of new optical technologies including cross-polarized optical coherence tomography, confocal laser endomicroscopy resulting high-resolution images of brain tissue, tumor and vascular network, etc. are intensively studied [31, 32, 33, 34, 35, 36].
Limitations of the study: short follow-up and small sample size, there were patients with gliomas of various malignancy grade including two patients after previous surgery in other institutions without subsequent adjuvant therapy. The question of possible vascular fluorescence in patients with low-grade gliomas and tumors with other histological and molecular genetic features also needs to be studied.
Authors’ participation:
Concept and design of the study — A.P., S.Ch., V.O., M.R., I.P., T.S., V.L., V.Ch.
Collection and analysis of data — P.N., V.O., S.G., A.K., A.M., D.Ch., A.B., G.D., T.S.
Statistical analysis - S.Ch., V.O., S.G., A.K., A.M., D.Ch., G.D.
Writing the text — A.P., V.O., S.G., A.K., A.M., D.Ch., M.V., N.Z., A.B., I.P., G.D., T.S., V.L., K.Ya., V.Ch.
Editing — A.P., S.Ch., P.N., V.O., S.G., A.K., A.M., D.Ch., M.R., N.Z., A.B., I.P., G.D., T.S., V.L., K.Ya., V.Ch.
This research was supported by the Russian Foundation for Basic Research (Grant 17-00-00158, “Analysis of clinical and molecular genetic patterns of human brain gliomas with a long period of overall survival”).
The authors declare no conflict of interest.
Commentary
The authors discuss an important question of the quality of resection of glioblastoma. 5-ALA fluorescence for intraoperative diagnosis of damaged vessels is described for the first time. This research confirmed the need for resection of fluorescent vessels considering the risk of malignant vascular invasion. According to literature data, vessels do not accumulate a fluorescent agent. Therefore, vascular accumulation of 5-ALA may be regarded as malignant invasion of vascular wall and perivascular tissues. Resection of the affected vessel (in case of anatomical and physiological capability) excludes one of the ways for tumor progression. New optical neuroimaging technologies are valuable to improve quality of resection of malignancies.
This study is unique and will be interesting and useful for neurosurgeons specialized in surgery for brain malignancies, as well as specialists using intraoperative fluorescence diagnosis in neurooncology.
V.A. Lazarev (Moscow, Russia)
Commentary
The authors’ report is devoted to description of vascular fluorescence in brain glioblastomas. The authors have advanced experience of fluorescence diagnosis in surgical treatment of various brain tumors. Glioblastomas are brightly fluorescent tumors with sensitivity of fluorescence up to 90—95% (Goryainov S.A., Potapov A.A., 2018, 2019). However, neoangiogenesis is often observed in the structure of tumors. The authors noted intraoperative fluorescence of tumor vessels during resection of glioblastomas.
A sample enrolled 14 patients. Glioblastoma was verified in 10 patients, anaplastic astrocytoma — in 4 cases. The authors found vessels with homogeneous or fragmentary fluorescence of the walls after resection of tumor in 4 out of 10 patients with glioblastoma. Vascular fluorescence was absent in patients with anaplastic astrocytoma. A thorough histological examination of infiltrated vascular walls was carried out.
This research makes it possible to identify intraoperative fluorescence of infiltrated vessels and demonstrates a possible pathway for tumor progression, i.e. malignant vascular wall infiltration followed by tumor spread along the vessels.
A.Kh. Bekyashev (Moscow, Russia)