Gross-total resection for brain tumors is currently recognized as a favorable prognostic factor for survival in patients with diffuse gliomas Grade II—IV [1]. At the same time, it is known that tumor cells spread beyond the boundaries of microsurgical visibility, contrast-enhanced area and/or zone of abnormal T2-signal in patients with diffuse gliomas [2—6]. Feasibility of more extensive “supratotal” resection is discussed for improvement of survival in patients with gliomas Grade II [7, 8]. Extensive resection of gliomas Grade III—IV is also discussed [9—11]. Similar questions are posed regarding resection of brain metastases. Depth of invasion in various morphological types of tumors remains unclear despite the apparent nodal growth of these neoplasms [12]. Resection of metastases with perifocal zone within 5 mm beyond the tumor has already become a standard [13]. However, even this approach is associated with recurrences [14]. Further studies of the effectiveness of the so-called “supramarginal” metastatic resection are required [12].
Modern methods for determining resection margins including metabolic navigation [15, 16] and intraoperative MRI [17, 18] can increase quality of resection. However, none of modern methods of intraoperative diagnosis guarantees total resection and absolute cytoreduction. Moreover, resection of tumor within its anatomical boundaries is associated with certain limitations if the neoplasm is adjacent to functionally significant zone. Indeed, patient’s functional status is not only a reliable predictor of survival, but also an essential factor of quality of life. The term “onco-functional balance” in modern neurosurgery of gliomas denotes this problem. Currently, this balance is achieved by intraoperative comparison of anatomical and “functional” boundaries of the tumor.
Some authors additionally analyze tumor boundaries regarding residual tumor tissue despite all advantages of modern imaging technologies [19]. In our study, we applied multifocal cytological examination of resection. Diagnostic value of this method for assessing CNS tumor spread has been demonstrated in various reports devoted to cytological [20, 21] and genetic [22, 23] verification of tumor tissue in patients with glial tumors [20—22] and cerebral metastases [20, 22, 23 ]. Sensitivity and specificity of cytological intraoperative examination are similar to the same values in case of histological examination using a cryostat [24].
Purpose — to analyze the possibilities of cytological study of resection margins of glial and metastatic tumors for objective assessment of resection quality.
Material and methods
The study included 35 patients (20 men and 15 women) with supratentorial brain tumors who underwent open surgery. There were 15 patients with brain metastases (melanoma — 5, carcinoma — 10), 13 patients with diffuse gliomas accumulating contrast agent (glioblastoma — 12 patients, glioma Grade III — 1 patient), 7 patients with diffuse gliomas and no contrast enhancement (Grade II — 5 patients, Grade III – 2 patients). Exclusion criteria: age <18 years, severe somatic diseases, suspected lymphoproliferative intracerebral process. All patients signed an informed consent. The ethics committee approved study protocol.
Primary surgery without previous adjuvant treatment was applied in all patients with non-contrast-enhancing diffuse gliomas. Redo surgery with radiotherapy between recurrences was performed in 4 out of 13 patients with contrast-enhancing gliomas Grade III—IV and 4 out of 15 patients with metastases. These patients were of particular interest regarding the effect of radiotherapy on perifocal presence of tumor cells.
All patients underwent preoperative contrast-enhanced MRI (Magnevist 0.2 ml/kg) and DTI-tractography of the white matter pathways (1.5 T Siemens Essenza MRI scanner). The data were compiled using the NordicBrainEx software. According to DTI-tractography data, corticospinal tract was involved in 24 out of 35 cases (68.6%), arcuate tract — in 11 out of 21 (52.4%) patients with tumor of dominant hemisphere. Neurophysiological monitoring was carried using cortical bipolar and subcortical monopolar stimulation (32-channel INOMED ISIS IOM scanner). Motor mapping was performed in 17 cases, speech cortical mapping — in 5 out of 35 cases.
The models of motor and associative (arcuate, oblique frontal) tracts were loaded into Medtronic S7 neuronavigation station. Standard surgical approaches and microsurgical resection of glial and metastatic brain tumors were used.
Histological and immunohistochemical examination of intraoperative specimens was used to confirm the final diagnosis. Resection of tumor was additionally associated with multifocal (2—5 sites) sampling of tissues for cytological examination (volume ͌ 0.003 cm3). Specimens were sampled in the nearest to tumor perifocal zone and the most distant points at the border of extended resection. Time of cytological examination is about 15 min.
Definition of the boundaries of the so-called "supratotal" resection is quite subjective. In this study, extended resection included aspiration zone within at least 5 mm from the visible tumor edge. Samples were harvested from this area. Another sampling point was at the tumor edge identified visually through a microscope in combination with neuronavigation data. Specimens were labeled, centrifuged, placed on a glass slide, fixed and stained with MGG express dye (Bio-Optica, Italy). Stained specimens were examined (Nikon ECLIPSE NI-U light microscope) under 100-, 200- and 1000-fold oil immersion magnification. Cellularity, presence of tumor cells and other elements were analyzed under low magnification (x100, 200). Structural features of the nucleus, nuclear membrane and cytoplasm were evaluated under 1000-fold magnification. A total of 154 samples were examined. The data were recorded in the individual patient’s registration card.
Control contrast-enhanced MRI in early postoperative period was performed in 29 patients. Analysis of resection quality is shown in Table 1.

Statistical analysis was performed using the Statistica 6.1 software program. Diagnostic significance of cytological examination was compared with postoperative MRI using logistic regression analysis. Within-group analysis was carried out using Pearson test (c2). Differences were significant at p-value<0.05.
Results
We present data on perifocal tumor cells distribution in study groups.
Cerebral metastases
There were 64 samples in patients who underwent resection of metastases (2—5 per a patient). The following data were observed (Fig. 1).

Tumor cells in the perifocal zone was the most common finding in patients with carcinoma (5 out of 10) and melanoma (2 out of 5 samples). Tumor cells at the border of extended resection were less common (only 3 samples in 2 patients) while total resection was confirmed by MRI in 92% and 100% of cases.
Case report
A 70-year-old patient N. had a metastasis of melanoma in the right parietal lobe (Fig. 2a—c).

This case demonstrates one of the potential causes of recurrent tumors due to endovascular dissemination of tumor cells.
Contrast-enhancing gliomas Grade III—IV
There were 60 samples of contrast-enhancing gliomas Grade III—IV. Tumor cells in the nearest perifocal zone were found in 22 out of 32 samples (68.8%), at the border of extended resection — in 14 out of 20 (70%) samples (χ2=0.009, p=1). Tumor cells were verified in 7 out of 8 (87.5%) control samples (Fig. 3).

Tumor cells in the nearest perifocal zone were found in 12 out of 13 patients, at the final border of resection — in 11 out of 13 patients. MRI-based surgical radicalness was over 99% in both cases with cytologically negative margins. Tumor cells in the nearest perifocal zone and at the border of extended resection were detected in 11 patients. That, extended surgery did not improve quality of resection considering cytological data.
Non-contrast-enhancing diffuse gliomas Grade II—III
There were 33 samples in 7 surgeries for non-contrast-enhancing gliomas Grade II—III (4—5 per a patient).
Tumor cells in perifocal zone were detected in 10 out of 17 samples (58.9%), in supramarginal zone — in 4 out of 11 (36.4%) samples (χ2=0.009, p=1). Tumor was verified in 3 out of 5 (60%) control samples (Fig. 4).

Functionally justified limitation of resection is shown in the following observation.
A 51-year-old patient Kh. had left-sided oligoastrocytoma (Grade II, NOS) with primary localization in the middle frontal gyrus (Fig. 5g).

In general, higher quality of resection confirmed by MRI data was associated with less incidence of detection of tumor cells at the border of final resection in at least one of the samples (“cytological” radicalness) (B0=19.2; B1=-20.8; c2=13.6; n=29, p=0.001; P(+)=50% with a radicalness of 92.6%) (Fig. 6).

Similar models were created in each of the selected groups: cerebral metastases (B0=3951; B1=-4318; c2=7; n=12; p=0.008), non-contrast-enhancing gliomas (B0=263.8; B1=-330.5; c2=8.3; n=6, p=0.004) and contrast-enhancing gliomas (B0=61 B1=-61; c2=3.6; n=11, p=0.056).
Detection of tumor cells was associated with a large volume of contrast-enhancing tumor in accordance with preoperative MRI in the groups of cerebral metastases and contrast-enhancing gliomas (B0=-1.1; B1=-0.04; c2=6.35; n=28, p=0.012) (Fig. 7).

Redo surgery and previous radiotherapy were not associated with “cytological” radicalness (c2=0.5, p=0.704).
Discussion
Primary data showed that cytological examination of perifocal zone was quite informative as an additional tool for assessing quality of resection of gliomas and metastases. Moreover, this technique was even more sensitive than postoperative MRI in some cases.
Extended resection of gliomas Grade III—IV has no advantages regarding “cytological” radicalness, because tumor cells were found in the nearest perifocal zone and at the border of extended resection with approximately the same incidence (68.8% and 70%, respectively).
At the same time, “supramarginal” resection of cerebral metastases was associated with less incidence of detection of tumor cells at the border of extended resection compared with the nearest perifocal zone (10.3% vs. 28.6%). Therefore, this method may be justified. Undoubtedly, further trials are required because insignificant differences (p=0.102) may be due to small sample size.
Conclusion
The most significant findings of multifocal cytological examination of resection margins were the data on the incidence of tumor cells detection within various distances from the tumor node. These results ensure certain pros and cons in discussion about the need for extended resection of neoplasms. Intraoperative on-line cytological examination to clarify surgical strategy seems perspective. Significance of this information for prediction of recurrences, clarifying the boundaries of target for radiotherapy and chemotherapy strategy remains debatable
The authors declare no conflicts of interest.