Quality indicators for evaluating the 30-day postoperative outcome in pediatric brain tumor surgery: a 10-year single-center study and systematic review of the literature

OBJECTIVE

Surgery is the cornerstone in the management of pediatric brain tumors. To provide safe and effective health services, quantifying and evaluating quality of care are important. To do this, there is a need for universal measures in the form of indicators reflecting quality of the delivered care. The objective of this study was to analyze currently applied quality indicators in pediatric brain tumor surgery and identify factors associated with poor outcome at a tertiary neurosurgical referral center in western Norway.

METHODS

All patients younger than 18 years of age who underwent surgery for an intracranial tumor at the Department of Neurosurgery at Haukeland University Hospital in Bergen, Norway, between 2009 and 2020 were included. The primary outcomes of interest were classic quality indicators: 30-day readmission, 30-day reoperation, 30-day mortality, 30-day nosocomial infection, and 30-day surgical site infection (SSI) rates; and length of stay. The secondary aim was the identification of risk factors related to unfavorable outcome. The authors also conducted a systematic literature review. Articles concerning pediatric brain tumor surgery reporting at least two quality indicators were of interest.

RESULTS

The authors included 82 patients aged 0-17 years. The 30-day outcomes for unplanned reoperation, unplanned remission, mortality, nosocomial infection, and SSI were 9.8%, 14.6%, 0%, 6.1%, and 3.7%, respectively. Unplanned reoperation was associated with eloquent localization (p = 0.009), primary emergency surgery (p = 0.003), and CSF diversion procedures (p = 0.002). Greater tumor volume was associated with unplanned readmission (p = 0.008), nosocomial infection (p = 0.004), and CSF leakage (p = 0.005). In the systematic review, after full-text screening, 16 articles were included and provided outcome data for 1856 procedures. Overall, the 30-day mortality rate was low, varying from 0% to 9.3%. The 30-day reoperation rate varied from 1.5% to 12%. The SSI rate ranged between 0% and 3.9%, and 0% to 17.4% of patients developed CSF leakage. Four studies reported infratentorial tumor location as a risk factor for postoperative CSF leakage.

CONCLUSIONS

The 30-day outcomes in the authors' department were comparable to published outcomes. The most relevant factors related to unfavorable outcomes are tumor volume and location, both of which are not modifiable by the surgeon. This highlights the importance of risk adjustment. This evaluation of quality indicators reveals concerns related to the unclear and nonstandardized definitions of outcomes. Standardized outcome definitions and documentation in a large and multicentric database are needed in the future for further evaluation of quality indicators.

P14.62 Overall survival from glioblastoma: partial resection versus biopsy

BACKGROUND

Maximal safe resection is standard of care in patients with glioblastoma. Partial resection or biopsy are alternative surgical approaches when macroscopic complete resection is unachievable. Survival benefit from partial resection remains uncertain. We aimed to evaluate overall survival from glioblastoma in patients having undergone partial resection compared to biopsy.

MATERIAL AND METHODS

We retrospectively identified all patients with histologically confirmed glioblastoma having undergone partial resection or biopsy in Western Norway between 1.1.2007 and 31.12.2014. Clinical characteristics and radiology reports were extracted from electronic medical records. Categorical data were compared by chi square test or Fishers exact test, and continuous data by non-parametric tests. Kaplan Meier method and log rank test were used for survival analyses.

RESULTS

We identified 158 patients diagnosed with glioblastoma and having undergone biopsy or partial resection. Biopsy was performed in 52 patients (32.9%) and partial resection in 106 patients (67.1%). Median age (range) was 62.5 (18.1–82.3) in the biopsy group and 62.2 (27.9–85.1) in the partial resection group (p=0.90). Median Charlson comorbidity score was four in both groups. Multifocality was observed in 46.2% of patients in the biopsy group, compared to 27.4% of patients in the partial resection group (p=0.02). Deep-seated tumour localisation was also more frequent in the biopsy group than in the partial resection group, seen in 17.3% vs 5.7% of the patients (p=0.04). There was no difference in chemoradiotherapy (CRT) treatment between the groups. CRT according to the Stupp protocol, less intensive CRT and best supportive care was performed in 36.5%, 50.0% and 13.5% of patients in the biopsy group, compared to 45.3%, 49.1% and 5.7% in the partial resection group (p=0.20). Median overall survival in the biopsy group was 8.1 months (95% CI 5.2–11.1) compared to 11.1 months (95% CI 9.4–12.8) in the partial resection group (p=0.19). Median survival in the biopsy group was 13.8 months (95% CI 10.1–17.5), 6.5 months (95% CI 3.6–9.4), and 3.5 months (95% CI 0.0–7.7) for patients receiving CRT according to Stupp protocol, less-intensive CRT and best supportive care, respectively (p<0.001). The corresponding numbers in the partial resection group were 15.1 months (95% CI 13.2–16.9), 9.1 months (95% CI 7.5–10.6), and 1.5 month (95% CI 0.0–4.7) (p<0.001).

CONCLUSION

Median overall survival was slightly longer in patients having undergone partial resection compared to biopsy, however not statistically significant. Prospective studies are needed to evaluate the survival benefit from partial resection.

Treatment of chronic subdural hematoma by burr-hole craniostomy in adults: influence of some factors on postoperative recurrence

BACKGROUND

The study was conducted to determine the causative factors in the postoperative recurrence (PR) of chronic subdural haematomas (CSDHs) and to evaluate the efficacy of surgery in adults enrolled in this trial.

METHODS

99 patients with 121 CSDHs, who were operated on between January 1999 and December 2001, were studied. We evaluated the PR rate related to anamnestic, clinical, surgical and neuroradiological imaging variables. In addition, we reviewed the number and the type of repeated operations, complications of surgery and the outcomes at one, three and 12 months.

FINDINGS

82.6% of lesions were successfully treated following the initial evacuation, and 95.9% of lesions following a second procedure. The PR rate was 14.9%. A significantly high PR rate was found to be associated with separated type, frontal base type, a midline displacement >5 mm and the presence of acute subdural clots in cranial base type on CT scans obtained within four days postsurgery. The interval from head trauma to initial surgery <60 days, the maximum width of subdural space >10 mm and massive collection of air in the subdural space tended to give a high PR rate. The PR rate associated with the homogeneous type of CSDHs was significantly low.Age, sex, cause of CSDH, anticoagulant therapy, preoperative neurological presentation, concomitant disease, variables on preoperative CT scans, and surgical factors such as the extent of the surgical procedure, use of drainage, duration and volume of drainage were not significantly associated with PR rate.

CONCLUSIONS

It is important to identify factors leading to a high or a low PR rate in the treatment of CSDHs because this may help to select appropriate surgical procedures and postoperative management to treat this condition efficiently.

The glial precursor proteoglycan, NG2, is expressed on tumour neovasculature by vascular pericytes in human malignant brain tumours

Glial precursor cells express NG2 and GD3 in the developing brain. These antigens are both over-expressed during neoplasia, which suggests they may have specific functions in the malignant progression of human brain tumours. This study describes the expression of NG2 and GD3 in 28 paediatric and adult brain tumours. Glioblastoma biopsy spheroids were also implanted into nude rats to assess the regional distribution of the molecules within the tumour. These xenografts showed extensive infiltration and growth that mimicked the growth patterns of human gliomas in situ. NG2 was identified in 20 out of 28 brain tumours, where the expression was confined to the main mass of the tumour, and was reduced towards the tumour periphery. NG2 was mainly associated with blood vessels on both the pericyte and basement membrane components of the tumour vasculature. Ki67 (MIB-1) labelling indicated that NG2 expression was associated with areas of high cellular proliferation. Conversely, all the tumours expressed GD3, which was present both in the tumour main mass and throughout the periphery. Thus, the expression of NG2 may be indicative of tumour progression and might be an amenable target for future therapeutic interventions.

The glioma-associated gangliosides 3'-isoLM1, GD3 and GM2 show selective area expression in human glioblastoma xenografts in nude rat brains

This work describes the in vivo expression and distribution of glioma-associated gangliosides (GD3, GM2, 3'-isoLM1) in a novel human brain tumour nude rat xenograft model. In this model, the tumours, which are established directly from human glioblastoma biopsies, show extensive infiltrative growth within the rat brain. This model therefore provides an opportunity to study ganglioside expression not only within the macroscopic tumour, but also in brain areas with tumour cell infiltration. The ganglioside expression was studied by confocal microscopy of immunostained brain sections using antiganglioside monoclonal antibodies. Xenografts from four human glioblastoma multiformes were established in rats and the brains removed after 3-4 months. Ganglioside GD3 was expressed in the tumour parenchyma while ganglioside 3'-isoLM1 was more abundantly expressed in the periphery of the tumour associated with areas of tumour cell invasion. GM2 expression was only seen in one tumour, where it was located within the main tumour mass. Double staining with a pan antihuman monoclonal antibody (3B4) and the antiganglioside monoclonal antibodies confirmed that the ganglioside expression was associated with tumour cells. This work supports the concept of different biological roles for individual gangliosides and indicates that antibodies or ligands directed against GD3 and 3'-isoLM1 might be complementary when applied in the treatment of human glioblastomas.

Biological mechanisms of glioma invasion and potential therapeutic targets

The current understanding of glioma biology reveals targets for anti-invasive therapy which include manipulations of extracellular matrix and receptors, growth factors and cytokines, proteases, cytoskeletal components, oncogenes and tumor suppressor genes. A better understanding of the complex regulation and the signalling molecules involved in glioma invasion is still needed in order to design new and effective treatment modalities towards invasive tumor cells. Representative and valid in vitro experimental systems and animal models of gliomas are necessary for the characterization of the invasive phenotype and further development of anti-invasive therapy. In the future, it will probably be important to move from comparative genomic modelling through protein characterization based on advanced proteomic techniques to analyse tissue samples, where the aim for gliomas should be to compare invaded and non-invaded tissue. This will hopefully render promising new therapeutic targets for gliomas.

Local endostatin treatment of gliomas administered by microencapsulated producer cells

We describe a technique for the treatment of malignant brain tumors based on local delivery of the anti-angiogenic protein endostatin from genetically engineered cells encapsulated in ultrapure sodium alginate. Alginate consists of L-guluronic and D-mannuronic acid, which in the presence of divalent cations forms an extended gel network, in which cells reside and remain immunoisolated, when implanted into the rat brain. Here, we show that endostatin-transfected cells encapsulated in alginate maintain endostatin secretion for at least four months after intracerebral implantation in rats. During the implantation period 70% of the encapsulated cells remained viable, as opposed to 85% in in vitro-cultured capsules. Rats that received transplants of BT4C glioma cells, together with endostatin-producing capsules (0.2 microg/ml per capsule), survived 84% longer than the controls. The endostatin released from the capsules led to an induction of apoptosis, hypoxia, and large necrotic avascular areas within 77% of the treated tumors, whereas all the controls were negative. The encapsulation technique may be used for many different cell lines engineered to potentially interfere with the complex microenvironment in which tumor and normal cells reside. The present work may thus provide the basis for new therapeutic approaches toward brain tumors.

Laminin expression by glial fibrillary acidic protein positive cells in human gliomas

Extracellular matrix components are regarded as important substrates for invasive tumor cells. The present work focuses on the expression of laminin in the brain in response to invading brain tumors. Biopsies obtained from tissue macroscopically evaluated as the border zone between tumor and normal brain, in 5 patients undergoing surgery for glioblastoma multiforme, were examined by immunocytochemistry and scanning confocal microscopy for the expression of laminin and glial fibrillary acidic protein. Laminin was mainly found in all the specimens associated with the basal lamina of blood vessels, but a variable degree of punctate laminin deposits were also observed in the parenchyma not associated with blood vessels. In the specimens with substantial deposits, scanning confocal microscopy showed that some of the laminin co-localized with intracellular glial fibrillary acidic protein. Punctate deposits of laminin were also seen in an intracranial BT4C rat glioma model, where it was particularly abundant in the brain/tumor confrontation zone. Previous in vitro studies have shown that laminin, among several extracellular matrix components, represent a highly permissive substrate for glioma cell migration. The presented results indicate that laminin can be produced by glial fibrillary acidic protein positive cells during glioma cell invasion in humans. This glycoprotein may thus represent one important substrate among many, which contribute to the invasive phenotype of gliomas.

Extracellular matrix-induced cell migration from glioblastoma biopsy specimens in vitro

The present knowledge about the interaction between the extracellular matrix (ECM) and gliomas is mostly based on studies of permanent cell lines. Since such cultures have undergone an extensive clonal selection in vitro, the experimental results obtained may be quite different from those obtained from studies on true biopsy specimens. The present work demonstrates how different ECM components affect tumor cell migration from human glioblastoma specimens grown as biopsy sample spheroids. Biopsy specimens from 12 glioblastomas and 1 gemistocytic astrocytoma were included in this study. Spheroids were directly initiated from the biopsy specimens, and after 3-4 weeks in culture, they were used in a migration assay. A custom-made filtered medium, where the high molecular weight (>100 kDa) proteins were removed, was supplemented with the following ECM components: laminin, fibronectin, collagen type IV and vitronectin. The cell migration was negligible when spheroids were propagated in the filtered medium. The ECM components as well as complete DMEM evoked strong stimulatory effects on different biopsy specimens. Opposed to that observed earlier for permanent glioma cell lines, highly variable responses were observed between the different biopsy samples on the various ECM components. In general, correlation analyses revealed that specimens that were strongly stimulated by laminin were also stimulated strongly by fibronectin, collagen type IV and vitronectin. This suggests that the capacity to migrate as a response to ECM was confined more to each biopsy specimen than to any specific ECM component. Since biopsy sample spheroids, as original tumors, consist of different cell types, an immunohistochemical characterization of the migrating cells was also performed. Anti-glial fibrillary acidic protein (GFAP) staining revealed both GFAP-positive and -negative migrating cells. Immunostaining for von Willebrand factor and CD11b indicated that the migrating cells were neither endothelial nor microglial cells. This study, therefore, indicates that migratory responses of glioma biopsy specimens to different ECM components is much more heterogeneous than that observed earlier for cell lines. Furthermore, the presented findings support the notion that gliomas may utilize different cell surface receptors for their migration, depending on the cell substrates available.

Stimulation of extracellular matrix components in the normal brain by invading glioma cells

Malignant gliomas are characterized by an extensive invasion of tumor cells into the normal brain parenchyma. A substantial amount of data indicates that cell movement in general is regulated by specific interactions between extracellular matrix components and specific cell-surface receptors. In the present work, multicellular spheroids from 4 human glioma cell lines (U-373Mg, A-172Mg, U-251Mg and HF-66) were confronted with normal rat brain cell aggregates in vitro, which resulted in a progressive invasion of tumor cells into the brain aggregates. The co-cultures were then sectioned and immuno-stained for specific extracellular matrix components (laminin, fibronectin and collagen type IV) and for specific cell-surface receptors which bind to these components (integrins beta1, beta4, alpha3, alpha6). In addition, flow-cytometric measurements and Northern blot analyses showed expression of several different integrins within the cell lines. The alpha3 subunit was expressed strongly in all cell lines. Whereas the beta1 subunit was expressed weakly in exponentially growing monolayer cultures, it showed a pronounced expression in multicellular spheroids, indicating that the integrin expression may vary depending on the micro-environment within a tumor. Furthermore, normal brain tissue was able to produce laminin when confronted with the glioma cells, which also was observed for fibronectin and collagen type IV. The relevance of our observations to the in vivo situation was investigated further by immuno-staining 5 human glioma biopsy samples for laminin. In some areas of the tumors, specific deposits of laminin were observed. In conclusion, we have shown that normal brain tissue has the ability to produce extracellular matrix components, such as laminin, collagen type IV and fibronectin, when confronted with invading glioma cells. Our results show that the glioma cells express specific integrins which can interact with these extracellular matrix components. Such interactions may facilitate tumor cell migration and invasion.

Role of high molecular weight extracellular matrix proteins in glioma cell migration

Malignant human gliomas are characterized by an uncontrolled cell proliferation and infiltrative growth within the brain. Complete surgical removal is difficult due to disseminated tumour cells, and the fundamental mechanisms responsible for this spread are poorly understood. An extensive tumour cell movement along blood vessels is frequently observed and this may be due to specific interactions between tumour cell surface receptors and specific extracellular matrix (ECM) components present in conjunction with vascular elements. In order to investigate the influence of ECM on glioma cell migration, three different human glioma cell lines (U-373 MG, A-172 MG and HF-66) were exposed to known ECM components of the basement membrane (laminin, fibronectin and collagen type IV). Cell migration from multicellular spheroids was studied, using a custom-made medium which was prepared by removing the high molecular weight protein fraction (>100 kDa) from newborn calf serum by ultrafiltration. To this medium, the specific ECM components were added. For two of the cell lines (A-172 MG and U-373 MG), laminin was the most potent stimulator of glioma cell migration; the effect of laminin exceeded that evoked by ordinary serum-supplemented medium. For the HF-66 cell line, fibronectin was the most potent stimulator of migration. Western blot analysis showed that the A-172 MG and HF-66 cell lines expressed low amounts of laminin compared with U-373 MG, which showed extensive intrinsic synthesis of this ligand. U-373 MG was the only cell line that migrated in pure filtered medium. The cells stimulated by fibronectin expressed a different morphology from those stimulated by laminin suggesting that specific ECM-receptor binding may activate different cytoskeletal components within the cells. Furthermore, it was shown that there was no difference in the amount of protein synthesis between cells grown in filtered medium and in filtered medium supplemented with different ECM components. This suggests that ECM-induced cell migration is not dependent on a high level of protein synthesis. It is also shown that alpha3 integrin, which is a receptor-subunit for laminin, fibronectin and collagen type IV, was highly expressed in all cell lines. This study indicates that glioma cells need serum proteins with a molecular weight >100 kDa to migrate in vitro, and that laminin and fibronectin play an important role in this process.

Stimulation of glioma-cell migration by laminin and inhibition by anti-alpha3 and anti-beta1 integrin antibodies

An induction of laminin in the confrontation zone between tumor cells and normal brain tissue has been observed in our model systems in vivo and in vitro. In order to study the effects of ECM components on glioma-cell migration and invasion, we have used 2 lacZ-transfected glioma cell lines, AN1/lacZ and U-251 /lacZ. Cell migration from multicellular spheroids was studied using different types of media: DMEM with 10% serum, Ultra Culture medium, and filtrated DMEM with serum in which the protein fraction > 100 kDa had been removed by ultrafiltration. Laminin, fibronectin and collagen type-IV were individually added to the different media, and cell migration from the spheroids was studied. The results show that cell migration in both cell lines, was stimulated by laminin and fibronectin. Collagen type-IV stimulated only cell migration of U-251/lacZ cells. Scanning electron microscopy revealed an extensive change in cell shape as a result of laminin stimulation. Flowcytometric studies showed that both AN1/lacZ and U-251/lacZ strongly express the alpha3 beta1 integrin receptor, which can bind to several ECM components (laminin, fibronectin, collagen). Immunofluorescence microscopy demonstrated that the same integrin sub-units were expressed in multicellular spheroids. When monoclonal antibodies to alpha3 and beta1 were added to the laminin-stimulated cultures, cell migration was significantly reduced. This indicates that the alpha3 beta1 integrin receptor plays an important role during glioma-cell migration.

Migratory patterns of lac-z transfected human glioma cells in the rat brain

Malignant brain tumors are characterized by extensive tumor-cell infiltration into the normal brain tissue. The present work describes the migratory behavior of human glioma cells transplanted into the adult rat brain with the aim of exploiting the extent of active cell migration and passive cell displacement within the central nervous system. To detect every transplanted tumor cell, a stably bacterial beta-galactosidase (lac-z) transfected human glioma cell line was used. To distinguish between an active cell migration process and passive cell displacement, rat brains were also implanted with inert fluorescent polystyrene microspheres and the distribution of tumor cells and microspheres was studied 1 hr and 3 days after implantation. One hour after implantation the tumor cells were strictly localized at the implantation site. However, 3 days after implantation, both tumor cells and microspheres showed an extensive distribution within the brain. Confirming earlier neuropathological and experimental studies, it is shown that the lac-z-transfected glioma cells had the capacity to move within the Virchow-Robin and subarachnoid spaces. However, since fluorescent microspheres were also found in these areas, this spread of tumor cells may be primarily mediated by the extensive cerebrospinal fluid flow that exists within the brain. Three days after implantation, the glioma cells also showed an active migration over the corpus callosum. In comparison, the fluorescent microspheres showed only limited spread along the callosal body. It is concluded that the bacterial lac-z gene can be stably transfected into human glioma cells and, since every tumor cell can be visualized within the brain, this model provides a tool for studying the mechanisms behind tumor-cell invasion of the brain.