A Comparative Study of Squash Smear Cytology Diagnosis and Radiological Diagnosis with Histopathology in Central Nervous System Lesions.

Background: Space occupying lesions (SOLs) of central nervous system (CNS) constitutes important cause of neurological morbidity and mortality. Squash cytology is technically a simple and rapid intraoperative diagnostic tool. Radiology is supportive of histopathological diagnosis.
Objectives: To enumerate the histopathological patterns of various central nervous system (CNS) lesions, to correlate cytopathological diagnosis with histopathological diagnosis, and to correlate radiological diagnosis with histopathological diagnosis. Materials and
Methods: It was a retrospective study of CNS lesion cases from January 2015 to August 2018. Cytological-histopathological concordance and radiological-histopathological concordance were calculated. Chi-square test was the statistical tool used for statistical analysis.
Results: Histopathological diagnosis of 50 cases included neoplastic lesions (42 cases [84%]) and non-neoplastic lesions (8 cases [16%]). Correct diagnosis was achieved by squash cytology in 36 cases (72%) and radiological diagnosis in 25 cases (50%) by complete concordance. However, diagnostic accuracy of squash and radiology improved considerably by 90% and 76%, respectively, after applying partial concordance criteria. For the detection of neoplastic lesions, squash cytology had 98% and radiology had 80% diagnostic efficacy.
Conclusion: Preoperative radiological investigation and intraoperative squash smear cytology are complementary to each other. A multidisciplinary approach is necessary for the management of patients. Copyright:

INTRODUCTION

Space-occupying lesions (SOLs) of central nervous system (CNS) constitutes important cause of neurological morbidity and mortality.[12] Squash cytology has been established as a technically simple, rapid, inexpensive, fairly accurate, and dependable intraoperative diagnostic tool for diagnosing CNS lesions.[3] It guides the neurosurgeon during surgical resection and targeting the lesion.[4] It also helps the surgeon to monitor and modify the approach and plan the extent of surgery.[456]

Radiological investigation is supportive of an accurate histological diagnosis.[1] Investigative modalities like computed tomography (CT) and magnetic resonance imaging (MRI) are important preoperative diagnostic tools to detect CNS lesions.[2] The radiological diagnosis of an SOL, however, requires confirmation by histological examination of a tissue biopsy. Hence, the liaison between neurosurgeons, neuroradiologists, and neuropathologist is necessary for better management of the patient.[1] The current study was undertaken to enumerate histopathological patterns of various central nervous system lesions and to correlate cytopathological diagnosis and radiological diagnosis with histopathological diagnosis.

MATERIAL AND METHODS

The present study was conducted in the department of pathology from January 2015 to August 2018 for 3 years and 8 months. It was a retrospective observational type of study. The study was approved by the Institutional Ethics Committee. Sample size was calculated by using the formula for descriptive study. Fifty cases were analyzed. All cases presenting with CNS lesions, subjected to preoperative radiological evaluation, intraoperative cytopathological evaluation, and subsequent histopathological evaluation were included in the study. Those cases in which cytology smears were inadequate for interpretation and in which histopathological evaluation could not be performed were excluded from the study.

Preoperative radiological diagnosis was collected. Intraoperatively, tissue samples sent in isotonic saline were utilized to make smears for cytological evaluation by squash technique. Smears were prepared according to standard operating procedures.[45] Alcohol-fixed smears were stained routinely by hematoxylin and eosin (H&E), Papanicolaou (PAP) method, and May Grunwald Giemsa (MGG) stain. Special stains (Ziehl–Neelsen [ZN], Periodic acid-Schiff [PAS]) were used in selected cases. Tissue samples sent in 10% formalin were allowed to fix for 24 h. Gross examination was done and bits were given. Paraffin-embedded sections were routinely stained by H and E. Special stains (PAS, reticulin) were used in selected cases. Squash smear cytology was reported by a single pathologist. Corresponding histopathology was reported by two pathologists.

Cytomorphological diagnosis was compared with histopathological diagnosis and radiological diagnosis (CT scan or MRI) was compared with that of histopathology. When both CT scan and MRI diagnosis were available, MRI diagnosis was considered for concordance. Available clinical details and laboratory investigation data were also documented. All primary CNS tumors were histopathologically categorized according to the 2016 World Health Organization (WHO) classification.[7] Those lesions that were not included under the WHO classification were categorized as unclassified lesions.

Statistical analysis

Frequencies, Chi-square test, and crosstabs were used for calculation. Cytological diagnosis was compared with histopathological diagnosis and the diagnostic efficacy of squash cytology was evaluated. Radiological diagnosis was compared with histopathological diagnosis and the diagnostic efficacy of radiological investigation was evaluated. All statistical calculations were done through statistical software STATA version 13.

RESULTS

Fifty cases were analyzed. The CNS lesions ranged from 4 months to 85 years. Clustering of cases was seen in the fifth decade (mean = 42.7 years). Males were affected in the majority of cases (26 cases [52%]) with M:F ratio of 1.08:1. CNS lesions were more common in the intracranial region (42 cases [84%]) than the spinal cord region (8 cases [16%]). Temporal region (17 cases [34%]) was the most common intracranial site of involvement. Thoracic region (37 cases [74%]) was the most common site of involvement in the spinal cord region.

Out of 50 cases, 42 cases (84%) were neoplastic lesions constituting the majority and eight cases (16%) were non-neoplastic lesions. The distribution of various neoplastic and non-neoplastic CNS lesions and their diagnostic accuracies are enumerated in Table 1. Among the neoplastic lesions, diffuse astrocytic and oligodendroglial tumors (14 cases [28%]) constituted most common tumor. Embryonal tumors (1 case [2%]) and mesenchymal, nonmeningothelial tumors (1 case [2%]) constituted the least common tumors. Among the non-neoplastic lesions, inflammatory lesions (6 cases [12%]) constituted the most common lesion. Cystic (1 case [2%]) and vascular lesions (1 case [2%]) constituted least common lesions [Table 1].

Table 1

Diagnostic accuracy of various CNS lesions, WHO grades of primary CNS tumors, and their diagnostic accuracies by squash smear cytology and radiology

CNS Lesions
S.NO	Histopathology	Cases (n=50)	Squash smear cytology		Radiology
			DACC-S	DAPC-S	DACC-R	DAPC-R
I	Neoplastic lesions	42 [84%]	29 [69.0%]	37 [88.0%]	21 [50%]	33 [78.5%]
1	Diffuse astrocytic and oligodendroglial tumors	14 [28%]	7 [50%]	12 [85.7%]	8 [57.1%]	13 [92.8%]
	Diffuse astrocytoma	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
	Anaplastic astrocytoma	2 [4%]	1 [50%]	2 [100%]	0 [0%]	2 [100%]
	Gliosarcoma	1 [2%]	0 [0%]	1 [100%]	0 [0%]	1 [100%]
	Glioblastoma	5 [10%]	2 [40%]	3 [60%]	5 [100%]	5 [100%]
	Anaplastic oligodendroglioma	2 [4%]	2 [100%]	2 [100%]	0 [0%]	2 [100%]
	Oligoastrocytoma	2 [4%]	0 [0%]	2 [100%]	1 [50%]	1 [50%]
	Anaplastic oligoastrocytoma	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
2	Other astrocytic tumors	2 [4%]	0 [0%]	1 [50%]	0 [0%]	1 [50%]
	Pilocytic astrocytoma	1 [2%]	0 [0%]	1 [100%]	0 [0%]	1 [100%]
	Subependymal giant cell astrocytoma	1 [2%]	0 [0%]	0 [0%]	0 [0%]	0 [0%]
3	Ependymal tumors	2 [4%]	2 [100%]	2 [100%]	0 [0%]	1 [50%]
	Ependymoma	1 [2%]	1 [100%]	1 [100%]	0 [0%]	1 [100%]
	Anaplastic ependymoma	1 [2%]	1 [100%]	1 [100%]	0 [0%]	0 [0%]
4	Embryonal tumors	1 [2%]	1 [100%]	1 [100%]	0 [0%]	0 [0%]
	Embryonal tumor (NOS)	1 [2%]	1 [100%]	1 [100%]	0 [0%]	0 [0%]
5	Tumors of the cranial and parspinal nerves	4 [8%]	3 [75%]	4 [100%]	3 [75%]	3 [75%]
	Schwanomma	4 [8%]	3 [75%]	4 [100%]	3 [75%]	3 [75%]
6	Meningiomas	7 [14%]	7 [100%]	7 [100%]	6 [85.71%]	7 [100%]
	Meningothelial meningioma	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
	Fibrous meningioma	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
	Transitional meningioma	2 [4%]	2 [100%]	2 [100%]	1 [50%]	2 [100%]
	Psommomatous meningioma	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
	Angiomatous meningioma	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
	Metaplastic meningioma	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
7	Mesenchymal, non meningothelial tumors	1 [2%]	0 [0%]	0 [0%]	0 [0%]	0 [0%]
	Hemangioma	1 [2%]	0 [0%]	0 [0%]	0 [0%]	0 [0%]
8	Unclassified tumors	7 [14%]	6 [85.71%]	7 [100%]	2 [28.57%]	4 [57.14%]
	Chordoma	1 [2%]	1 [100%]	1 [100%]	0 [0%]	1 [100%]
	Esthesioneuroblastoma	1 [2%]	1 [100%]	1 [100%]	0 [0%]	0 [0%]
	Extra renal Wilms tumor	1 [2%]	0 [0%]	1 [100%]	0 [0%]	1 [100%]
	Plasma cell myeloma	2 [4%]	2 [100%]	2 [100%]	0 [0%]	0 [0%]
	Pituitary neoplasm	2 [4%]	2 [100%]	2 [100%]	2 [100%]	2 [100%]
9	Metastasis	4 [8%]	2 [50%]	2 [50%]	1 [25%]	3 [75%]
	Metastatic adenocarcinoma	3 [6%]	2 [66.66%]	2 [66.66%]	1 [33.3%]	3 [100%]
	Metastastic carcinoma: poorly differentiated	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
II	Non-neoplastic lesions	8 [16%]	7 [87.50%]	8 [100%]	4 [50%]	5 [62.50%]
1	Cystic	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
	Glio-ependymal cyst	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
2	Inflammatory lesions	6 [12%]	5 [83.33%]	6 [100%]	2 [33.33%]	3 [50%]
	Acute on chronic inflammatory lesion	1 [2%]	0 [0%]	1 [100%]	0 [0%]	1 [100%]
	Fungal granuloma	1 [2%]	1 [100%]	1 [100%]	0 [0%]	0 [0%]
	HIV encephalitis	1 [2%]	1 [100%]	1 [100%]	0 [0%]	0 [0%]
	Progressive multifocal leucoencephalopathy	1 [2%]	1 [100%]	1 [100%]	0 [0%]	0 [0%]
	Tuberculous lesion	2 [4%]	2 [100%]	1 [100%]	2 [100%]	2 [100%]
3	Vascular	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
	Infarct brain tissue	1 [2%]	1 [100%]	1 [100%]	1 [100%]	1 [100%]
	Total	50 [100%]	36 [72%]	45 [90%]	25 [50%]	38 [76%]
1	Grade I	13 [26%]	11 [84.61%]	12 [92.36%]	10 [76.92%]	11 [84.61%]
2	Grade II	4 [8%]	4 [100%]	4 [100%]	2 [50%]	3 [75%]
3	Grade III	6 [12%]	6 [100%]	6 [100%]	0	4 [66.66%]
4	Grade IV	7 [14%]	6 [85%]	6 [85%]	5 [71.4%]	6 [85.7%]
	Total	30 [60%]	27 [90%]	28 [93.3%]	17 [56.6%]	24 [80%]

DACC-S: Diagnostic Accuracy by Complete Concordance on squash; DAPC-S: Diagnostic Accuracy after Considering Partial Concordance on Squash; DACC-R: Diagnostic Accuracy by Complete Concordance on Radiology; DAPC-R: Diagnostic Accuracy after Considering Partial Concordance on Radiology. “Complete concordance” was applied to the cases in which intraoperative cytology or radiology diagnosis was exactly identical to the final histopathological diagnosis. “Partial concordance” in cytology was applied to cases in which ±1 grade of deviation in tumor grading was registered or when the diagnosis of the cell line of origin was not possible. “Partial concordance” in radiology was applied to cases in which radiological diagnosis showed minor deviation from final histopathological diagnosis but the lesion belonged to the same category or one of the radiological differential diagnoses matches with final histopathological diagnosis “Discordance” was considered in cases where intraoperative cytology diagnosis or preoperative radiology diagnosis differed from the final histological diagnosis

For squash smear cytology [Figures 1 and 2], the diagnostic accuracy was 72% by complete concordance and increased to 90% on applying partial concordance criteria. Neoplastic lesions showed better concordance than non-neoplastic lesions on applying complete concordance criteria. But non-neoplastic lesions showed better concordance than neoplastic lesions on applying partial concordance criteria. Among neoplastic lesions, the ependymal tumors, embryonal tumors, tumors of the cranial and paraspinal nerves, meningiomas, unclassified tumors, and diffuse astrocytic, and oligodendroglial tumors showed better concordance than other astrocytic tumors and metastatic tumors. But mesenchymal, nonmeningothelial tumors showed discordance. Among non-neoplastic lesions, all categories of lesions showed good concordance [Table 1].

Figure 1

(a-d) Plasma cell myeloma: (a) CT scan image (sagittal view): multiple punched-out lesions in lumbar vertebra. (b) Squash cytology. [PAP, 400×]. (c) Histopathology. [H&E, 400×]. (d) Immunohistochemistry: neoplastic cells strongly positive for lambda light chain. Inset: negative for kappa light chain. [IHC, 400×]

Figure 2

(a-d) Ependymoma and chordoma: (a) Squash cytology of ependymoma. Inset: Ependymal tumors cells. [H&E, 400×]. (b) Histopathology of ependymoma: ependymal canal. Inset: Ependymal pseudorosette. [H&E, 400×]. (c) Squash cytology of chordoma. [MGG, 400×] Inset: Polygonal tumors cells. [H&E, 400×]. (d) Histopathology of chordoma. [H&E, 400×]

For radiological diagnosis, the diagnostic accuracy was 50% by complete concordance and increased to 76% on applying partial concordance criteria. Neoplastic lesions showed better concordance than non-neoplastic lesions after applying partial concondance criteria. Among neoplastic lesions, the meningiomas, diffuse astrocytic and oligodendroglial tumors, tumors of the cranial and paraspinal nerves, and metastasis showed better concordance than other astrocytic tumors, ependymal tumors, and unclassified tumors. But embryonal tumor and mesenchymal, nonmeningothelial tumors showed discordance. Among non-neoplastic lesions, cystic lesions and vascular lesions showed better concordance than inflammatory lesions [Table 1].

Primary CNS neoplasms were categorized according to WHO grades and their diagnostic accuracies are enumerated in [Table 1]. WHO grade I tumors constituted the most common primary neoplastic CNS lesion. WHO grade II tumors constituted the least common primary neoplastic lesion. For squash smear cytology, the diagnostic accuracy was 90% by complete concordance and increased to 93.3% on applying partial concordance criteria. Concordance was better in WHO grade II and WHO grade III than WHO grade I and WHO grade IV. For radiological diagnosis, the diagnostic accuracy was 56.6% by complete concordance and increased to 80% on applying partial concordance criteria. Concordance was better in WHO grade IV and WHO grade I than WHO grade II. Concordance was relatively poor in WHO grade III. [Table 1]

The statistical values were calculated by Galen and Gambino method. For the detection of neoplastic lesions, the squash smear cytology had 41 true-positive cases, eight true-negative cases, one false-negative case, zero false-positive cases, sensitivity of 100%, specificity of 88.9%, false-positive error rate of 0%, false-negative error rate of 2%, positive predictive value of 97.6%, negative predictive value of 100%, and diagnostic efficacy of 98%. The P value, determining the efficacy of squash cytology for detecting CNS neoplasms was statistically highly significant (P < 0.001). For the detection of neoplastic lesions, radiological diagnosis had 37 true-positive cases, five false-positive cases, five false-negative cases, three true-negative cases, sensitivity of 88.1%, specificity of 37.5%, false-positive error rate of 10%, false-negative error rate of 10%, positive predictive value of 88.1%, negative predictive value of 37.5%, and diagnostic efficacy of 80%. The P value, determining efficacy of radiological diagnosis for detecting CNS neoplasms was statistically not significant (P < 0.070).

DISCUSSION

The term “intracranial space-occupying lesions (ICSOL)” is defined as any neoplasm, benign or malignant, primary or secondary, and any inflammatory or parasitic mass lying within the cranial cavity. It also comprises hematomas, different types of cysts, and vascular malformations.[89]

Intraoperative consultations about the pathology of CNS lesions are requested for a number of reasons. The reasons include confirming the presence of lesions, differentiating neoplastic from non-neoplastic, differentiating primary from secondary, estimating the degree (grade) of malignancy, determining tumor margins (e.g., meniningiomas), obtaining tissue for special purposes like culture, and ensuring safe handling and processing of tissue (e.g., Creutzfeldt–Jakob disease).[10]

Intraoperative squash cytological diagnosis is a fairly accurate, safe, simple, and reliable tool for the rapid diagnosis of CNS lesions. It is a preferred method as it offers great detail of cellular morphology by avoiding distortion and ice artifacts, which are often introduced by frozen sections. It helps neurosurgeons to plan the extent of surgery and modify it accordingly.[11] Radiological investigation is the only means for preoperative assessment of CNS lesions. Radiology helps to localize the lesion and plan the surgery. Histopathology is considered the gold standard investigation for the evaluation of CNS lesions.[2] The present study emphasizes the correlation of cytopathological diagnosis and radiological diagnosis with histopathological diagnosis.

Dogar et al[1], Govindaraman et al[4], Deshpande et al[11], Goyani et al[12] observed that most of the cases were seen in fourth decade in their study. Acharya et al[13], Kaki et al[14] documented that CNS lesions were common in the sixth decade. In contrast to the other studies, Das et al[15] and the present study showed clustering of cases in the fifth decade in their study.

Patil et al[3], Govindaraman et al[4] and Kaki et al[14] observed female predominance in their study. In contrast, Dogar et al[1], Jindal et al[6], Jindal et al[9], Deshpande et al[11], Goyani et al[12] and the present study showed male predominance.

Nanarng et al[16] documented fronto-temporal region as a predominant site of involvement in their study. Parietal region was the most common site of involvement in the studies conducted by Dogar et al[1], Goyani et al[12] and Kaki et al[14], In contrast to other studies, temporal region was the commonest site of involvement in the present study.

The distribution pattern of various CNS lesions in the present study was compared with other studies. [Table 2] In most of the studies, neoplastic lesions were common than non-neoplastic lesions. Among neoplastic lesions, diffuse astrocytic and oligodendroglial tumors were common in the present study. Patil et al[3], Govindaraman et al[4], Das et al[15], Acharya et al[13], Nanarng et al[16] and de Souza Balsimelli et al[17] and Jindal et al[18] documented similar observations in their studies. In contrast, Kang M et al[19] found meningeal tumors and Jindal et al.[6] found embryonal tumors as the commonest neoplastic lesion in their study, respectively. Jindal et al.[6] had done a study on the pediatric population in their study. Among non-neoplastic lesions, inflammatory lesions were the most common lesion in most of the studies [Table 2].

Table 2

Comparison of distribution of CNS lesions in various studies

Authors	Cases	Diffuse Astrocytic And Oligodendroglial Tumors	Other Astrocytic Tumors	Ependymal Tumors	Embryonal Tumors	Tumors Of The Cranial And Paraspinal Nerves	Meningiomas	Mesenchymal, Nonmeningothelial Tumors	Unclassified Tumors	Metastasis	Non-Neoplastic Lesion
Patil et al.[3] [2016]	50	16 [32%]	0	0	0	12 [24%]	13 [26%]	1 [2%]	0	1 [2%]	0
Govindaraman et al.[4][2018]	75	23 [30.66%]	2 [2.66%]	0	0	14 [18.66%]	18 [24%]	3 [4%]	0	4 [5.35%]	6 [8%]
Sunila et al.[5] [2017]	100	31 [31%]	5 [5%]	3 [3%]	4 [4%]	15 [15%]	22 [22%]	0	2 [2%]	5 [5%]	10 [10%]
Jindal et al.[6] [2017]	150	25 [16.66%]	33 [22%]	20 [13.33%]	36 [24%]	3 [2%]	7 [4.6%]	5 [3.33%]	0	0	0
Das et al.[15] [2018]	73	28 [38.35%]	0	2 [2.73%]	4 [5.47%]	10 [13.69%]	21 [28.76%]	3 [4.10%]	0	3 [4.10%]	0
Acharya et al.[13] [2016]	220	78 [35.13%]	1 [0.45%]	3 [1.35%]	11 [4.95%]	9 [4.05%]	62 [27.92%]	5 [2.25%]	0	11 [4.95%]	6 [2.70%]
Nanarng et al.[16] [2015]	75	24 [32%]	0	1 [1.33%]	0	6 [8%]	16 [21.33%]	0	3 [4%]	5 [6.66%]	16 [21.33%]
de Souza Balsimelli et al.[17] [2019]	113	37 [32.74%]	26 [23%]	3 [2.65%]	8 [7.07%]	2 [1.76%]	5 [4.42%]	4 [3.53%]	0	13 [11.5%]	14 [12.38%]
Jindal et al.[18] [2017]	150	39 [26%]	9 [6%]	9 [6%]	10 [6.66%]	22 [14.66%]	20 [13.33%]	4 [2.66%]	12 [8%]	8 [5.33%]	8 [5.33%]
Kang et al.[19] [2019]	454	64 [14.09%]	13 [2.86%]	10 [2.20%]	0	49 [10.79%]	98 [21.58%]	3 [0.66%]	89 [19.6%]	52 [11.45%]	36 [7.92%]
Present study	50	14 [28%]	2 [4%]	2 [4%]	1 [2%]	4 [8%]	7 [14%]	1 [2%]	7 [14%]	4 [8%]	8 [16%]

In the present study, diagnostic accuracy by complete concordance (DACC) and diagnostic accuracy after applying partial concordance criteria (DAPC) were calculated. Most of the studies had not clearly specified the criteria to be followed to consider concordance. This may be important because, not all the cases diagnosed by cytology and/or radiology can perfectly match with histopathological diagnosis, which is considered as the gold standard.

In the present study, on cytological diagnosis complete concordance was 72%, which has increased to 90% on application of partial concordance. Kang et al[19] documenteda similar experience in their study. [Table 3] Squash smear diagnosis was compared with histopathology with respect to WHO grades of primary CNS neoplasms. Good concordance was seen in WHO grade II and grade III tumors. Sunila et al.[5] made similar observations in their study. Sampling errors are known to affect the grading of glial tumors.[5]

Table 3

Comparison of diagnostic accuracies of cytological and radiological diagnosis in various studies

Comparison of diagnostic accuracy of cytological diagnosis
Authors	Cases	Complete concordance	Partial concordance
Patil et al.[3] [2016]	50	92%	–
Govindaraman et al.[4] [2018]	75	90.67%	–
Sunila et al.[5] [2018]	100	73.11%	84.94%
Jindal et al.[6] [2017]	150	94.67%	–
Das et al.[15] [2018]	73	84.93%	–
Acharya et al.[13] [2016]	220	83.78%	–
Nanarng et al.[16] [2015]	75	89.2%	–
de Souza Balsimelli et al.[17] [2019]	113	85.7%	–
Jindal et al.[18] [2017]	150	94.0%	–
Kang et al.[19] [2019]	454	76.2%	87.2%
Present study	50	72%	90%
Comparison of diagnostic accuracy of radiological diagnosis
Dogar et al.[1] [2015]	102	49.02%	–
Sunila et al.[2][2018]	88	45%	85%
Jindal et al.[9] [2016]	80	84.2%	–
Goyani et al.[12] [2015]	70	98.57%	–
Kaki et al.[14] [2017]	50	88%	–
Present study	50	50%	76%

Out of 50 cases, five cases (10%) were misdiagnosed (discordant) on cytology and 12 cases (24%) on radiology. Misdiagnosed cases and probable causes on cytology and radiology are summarized in Table 4. Reasons for misdiagnosis on cytology were mainly due to reactive cells, nonrepresentativeness of the sample, morphological similarity of cells, poor cellularity, and increased vascularity. Patil et al.,[3] Govindaraman et al.,[4] Nanarng et al[16], de Souza Balsimelli et al[17] and Mitra et al[20] had also misdiagnosed various CNS lesions and analysed pitfalls in their study. Deshpande et al[11] has documented that increased fibrous component, biopsies from cyst wall, inflammation and necrosis obscuring morphology, lack of architecture on cytology, reactive changes, and resistance to desegregation were the common causes for inconclusive cases in their study.

Table 4

Discordant cases of CNS lesions on cytology and radiology

Discordant cases of CNS lesions on cytology
Histopathology	Discordant cases	Cytology diagnosis	Probable cause for misdiagnosis
Glioblastoma	2	Glial neoplasm- low grade	Nonrepresentative sample
Subependymal giant cell astrocytoma	1	High-grade glial neoplasm	Pleomorphic cells and morphological similarity
Hemangioma	1	Chronic inflammatory lesion or giant cell lesion	Poor cellularity and increased vascularity
Metastatic adenocarcinoma	1	Meningioma	Morphological similarity
Discordant cases of CNS lesions on radiology
Histopathology	Discordant cases	Radiology diagnosis	Probable cause for misdiagnosis
Oligoastrocytoma	1	Periventricular glioblastoma	Similar radiological features
Subependymal giant cell astrocytoma	1	Glioblastoma	Similar radiological features
Anaplastic ependymoma	1	Metastasis	Edema and necrosis
Embryonal tumor (NOS)	1	Meningioma	Lobulated nature of the lesion with mass effect
Schwannoma	1	Meningioma	Similar radiological features
Hemangioma	1	Lytic lesion	Similar radiological features
Esthesioneuroblastoma	1	Lymphoma	Both present as hyperdense masses
Plasma cell myeloma	2	Tuberculosis Pott’s spine	Similar radiological features
Fungal granuloma	1	Glioblastoma	Similar radiological features
HIV encephalitis	1	Glioblastoma	Similar radiological features
Progressive multifocal leukoencephalopathy	1	Glioblastoma	Similar radiological features

Consistency of the tissue affects the diagnostic accuracy of the lesion. Soft and friable tissues can be easily smeared and yield good cellularity. Majority of gliomas, pituitary neoplasms, medulloblastomas, and metastatic carcinomas usually yield good cellularity and posed few diagnostic problems. Disadvantage of the smear technique lies with those lesions that are too firm to smear.[5] In the present study, squash smear cytology showed low specificity and high sensitivity. Mouriquand et al suggested multiple biopsies from different areas of the lesion may decrease false positivity and false negativity.[5] Accuracy of diagnosis on cytology depends upon familiarity with clinical history, tumor consistency, tumor location, differential diagnosis in a particular location, and cytohistological appearance of the potential etiologies of CNS lesions. It also depends upon the knowledge, interest, enthusiasm, and experience of a neuropathologist.[5]

In the present study, on radiological diagnosis complete concordance was 50%, which has increased to 76% on application of partial concordance. Jindal et al[9], Goyani et al[12] and Kaki et al[14] has documented higher concordance in their study. In contrast, Dogar et al.[1] and Sunila et al.[5] had observed lower concordance in their study.

Radiological diagnosis was compared with histopathology with respect to WHO grades of primary CNS neoplasms. Good concordance was seen in WHO grade I and grade IV tumors. Sunila et al.[2] documented better concordance in WHO grade II and WHO grade I in their study.

The reasons for misdiagnosis in radiology were mainly radiological similarity of the lesions and interfering factors like the presence of necrosis and edema. Tesfay et al.[8] concluded that perilesional edema of varying degrees may be seen in meningioma (extra-axial mass), which is also a feature of gliomas (more frequent in high-grade gliomas), metastases, and infections leading to misdiagnosis. Pant et al.[21] had also analyzed misdiagnosed cases in their study.

Pant et al.[21] had concluded that a location-based approach to CNS tumors on radiology would be helpful in establishing an appropriate differential diagnosis. Location-based algorithm in combination with specific imaging findings, such as the type of margins and signal, nature of contents (necrosis, hemorrhage, and calcification), pattern of enhancement, presence or absence of edema, mass effect, bony changes, and presence or absence of restriction on diffusion-weighted images, would permit a narrowed differential diagnosis. Kaki et al.[18] stated that MRI would remain the first-line investigation for diagnosing and evaluation of intracranial space-occupying lesion with a reasonable degree of diagnostic accuracy and with the advent of newer modifications of MRI such as MR spectroscopy and newer techniques like MR perfusion. Radiological investigation also showed low specificity and better sensitivity. Radiologically, similarity of the lesions poses difficulty in diagnosing the lesions accurately. Hence, radiological differentials can be offered to minimize the diagnostic difficulty.

In the present study, the sample size was relatively less in comparison with other studies. This was because both squash smear cytology and radiological investigation were considered to evaluate each case in the present study.

Radiological investigation done preoperatively helps the neurosurgeon to plan the surgery. Intraoperative squash smear cytology helps the surgeon to modify the extent of surgery.

Preoperative radiological investigation and intraoperative squash smear cytology are complementary to each other. Both squash smear cytology and radiological investigations would have their own merits and demerits. In the present study, squash smear cytology showed low specificity and high sensitivity. Similarly, radiological investigation also showed low specificity and better sensitivity. Squash cytology can be considered as a mirror image to histopathological diagnosis. Hence, it may be suggested that a multidisciplinary approach is necessary for the better management of the patients.

Acknowledgments

We sincerely thank cytopathology technicians, Department of Pathology for the kind co-operation extended to us for the work up of the cases during the course of the study.

Presentation

The research work was presented under PG Award paper category at 3rd Annual State level conference (IAPM AP Chapter) held from 08.09.2018 to 09.09.2018 at MIMS, Nellimarla, Vizianagaram, Andhra Pradesh, India.

Ethical Committee Approval

Ethical approval was obtained from Institutional Ethical Committee.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.