Comparison of Frozen Section and Squash Cytology as Intra-Operative Diagnostic Tool in Pediatric CNS Tumors
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.344656
Source of Support: None, Conflict of Interest: None
Keywords: Frozen section, pediatric CNS tumours, squash cytology
After liquid tumors, brain tumor accounts for the largest number of cancer deaths in children. Critical decisions regarding management including the extent of surgical resection depends on the accurate intra-operative diagnosis of central nervous system (CNS) tumors especially in children as dissection is difficult and requires more time and skill. Intra-operative diagnosis especially using a combination of squash and frozen section (FS) can help in accurate diagnosis in most cases.,,,,, Many a times even the assurance of the fact that lesional tissue has been obtained is enough for the neurosurgeons to proceed with their treatment modality.
Pediatric CNS tumors are unique due to their varied histological spectrum as compared to adults. Within the pediatric CNS tumors, the infantile tumors show even more diverse morphological spectrum as compared to older children. Hence, familiarity with these tumors is valuable in arriving at an accurate intra-operative diagnosis. Although FS is a well recognized and a useful adjunct in diagnosis of CNS tumors, there is limited literature available on utility of FS exclusively in pediatric cohort.
We have earlier discussed the role of intraoperative diagnosis using squash cytology alone in both adults and pediatric age group., Here, we have evaluated the diagnostic accuracy of FS and squash cytology along with the degree of correlation between these two procedures keeping histopathology as the gold standard in the diagnosis of pediatric CNS tumors.
This study was conducted over a period of 2.5 years in the department of Pathology in collaboration with the Neurosurgery department. A total of 55 cases of CNS tumors in patients up to 18 years of age were included in this study after clearance from the institutional ethical committee. Detailed clinical and radiological profiles of the patient were available during intra-operative consultation. A small portion of the specimen received was squashed using two clean glass slides and smeared. These slides were stained using hematoxylin and eosin (H and E) stain. Rest of the tissue was sent for FS using a cryostat. The section were picked up on a glass slide and stained with H and E stain. Rest of the tissue or more tissue sent later was put through tissue processing and final histopathological diagnosis was made on formalin-fixed paraffin-embedded (FFPE) tissue.
Squash cytology and FS were seen by two different trained pathologists initially in a blinded fashion. The diagnosis of squash and FS were noted separately. In case of discrepancy between the two, the findings of squash and frozen were reviewed together along with clinicoradiological findings and a final IOD was communicated. Histopathological diagnosis was considered as gold standard and the squash as well as FS interpretation were compared with it. Four groups were formed depending on the degree of correlation. Category I, II, and III indicated complete correlation, partial correlation, and no correlation between the intra-operative diagnosis (squash/FS) and FFPE. Complete correlation implies total agreement between FS/squash cytology and FFPE report. Partial correlation implied discrepancy in cytological grade within the same family of tumors. Category IV included unclassifiable tumors. A 2 × 2 contingency table was made and sensitivity, specificity, positive predictive value, negative predictive value, P value, and kappa value were calculated using GraphPad Prism.
A total of 55 patients were included in our study. Most of the pediatric patients with CNS lesions belonged to the age group of 6–15 years (60%) with a mean age of 9.76 ± 5.2 years. There was no significant sex preponderance. Supratentorial lesions (47.2%) were most common followed by the infratentorial (36.4%) and extra-axial lesions (16.4%). [Table 1] elaborates the clinicopathological profile of our patients.
Neoplastic lesions (96.4%) were predominant in our study. The common lesions encountered in our study were pilocytic astrocytoma (12.7%) followed by craniopharyngioma (10.9%), medulloblastoma (9.1%), and schwannoma (7.2%). Other tumors included were ependymoma, chordoma, central neurocytoma, diffuse astrocytoma, pleomorphic xanthoastrocytoma (PXA), and germinoma. Non-neoplastic lesions included a case of tuberculoma and non-specific inflammation.
Out of 55 patients, 35 (63.6%) showed complete correlation on squash, however, the correlation was higher with FS (74.5%), whereas the category of no correlation showed a higher percentage with squash (21.8%) as compared to FS (16.3%). High-grade gliomas, medulloblastoma, and craniopharyngiomas showed 100% correlation by both squash and FS Tumors like hemangiopericytoma showed 100% correlation by FS, whereas, had lesser accuracy using squash cytology. A case of germinoma was also diagnosed as small round cell tumor on squash but diagnosed accurately on frozen.
Rhabdoid meningioma, a high-grade meningioma in an adolescent, was diagnosed as atypical meningioma by both the procedures. Two cases of PXA (Grade II) were misdiagnosed as high grade glioma both on squash and FS. [Table 2] summarizes the correlation between squash, frozen, and histopathology in details. [Figure 1],[Figure 2],[Figure 3]
[Table 3] and [Table 4] show agreement between squash smear, FS, and histopathological grade group. The tumors were graded according to the updated 2016 WHO classification of CNS tumors. It did not show any significant difference between cytology and FS grade when low grade (Grade I, II) and high grade (Grade III, IV) tumors were considered together. WHO grading system was not applicable for 18.2% of the tumors.
Although the specificity and positive predictive value were comparable between the two methods, FS was seen to have a higher sensitivity (91.43%) as well as negative predictive value (75%). The two-sided P value was <0.0001 for both squash and frozen with likelihood ratio of 8.857 and 9.143, respectively.
We also calculated the kappa value as shown in [Table 5] to find out the degree of correlation between these two procedures. Kappa value in our study revealed that there was substantial agreement (0.7289) between the two procedures, thereby, implying that any single one can be used for intra-operative diagnosis if the other is not available or possible due to scanty amount of tissue received.
Accurate intra-operative diagnosis of pediatric brain tumors is important as they have different histological spectrum and location as compared to adult brain tumors. Jain et al. illustrated the histological profile of pediatric brain tumors in India in their multi-institutional study. Astrocytic tumors (34.7%), medulloblastoma, supratentorial primitive neuro-ectodermal tumors (22.4%), craniopharyngioma (10.2%), and ependymomas (9.8%) formed majority of the cases. In other series, it was shown that unlike adults, pilocytic astrocytoma, diffuse astrocytoma, embryonal tumors, and ependymomas forms around 70% of the pediatric CNS tumors, whereas meningiomas and schwannomas have lower incidence.,
There is sparse literature available comparing the role of both FS and squash preparation in CNS lesions, more so in the pediatric age group., A combination of cytological preparation along with FS analysis helps to bring out the striking features of different tumors that help to form a definitive intra-operative diagnosis., These two procedures are complimentary to each other.
Radioimaging findings which include location and extent of the tumor are extremely important pre-analytical factors. Chand et al. in their study have described the errors, limitations, and pitfalls in diagnosing CNS tumors intra-operatively in adults. In case of pediatric CNS tumors, in addition to site and location, the age at presentation is also very important; whether the patient is a neonate, infant, toddler, older child, or adolescent. In case of newborn, teratoma and other germ cell tumor remains the first differential while interpreting a case. But the scenario, changes in the toddler where ependymoma and medulloblastoma comes to the forefront. At that time, the accurate diagnosis becomes very significant as the extent of resection varies accordingly. As age progresses, the frequency of pilocytic astrocytoma and PXA increases gradually.
In our study, we found the specificity and positive predictive value of the two methods to be comparable, although FS had a higher sensitivity (91.43%) than squash cytology. Shah et al. in their study of 183 CNS tumors in adults found the diagnostic accuracy of FS (90.4%) to be slightly greater than squash preparation (89.7%) and concluded both the diagnostic tools to be comparable. In our study, we found substantial agreement (k = 0.7289) between the two procedures, thereby implying that any one of them can be used when the tissue received is scanty.
Various non-glial tumors found in our study were excluded from our statistical analysis as they could not be assigned any grade in WHO classification. For certain non-glial tumors like germinoma, a diagnosis of small round cell tumor (SRCT) was made, however, frozen section were extremely helpful in arriving at a diagnosis this case. A case of yolk sac tumor was reported as high grade glioma by both the procedures. IOD of germ cell tumors helps the neurosurgeon to modify the extent of resection as they are highly radiosensitive tumors and extensive resection of these deep seated tumors may lead to damage to adjacent vital structures.
Non-Hodgkin lymphoma was correctly diagnosed both on squash and FS, which avoided gross total resection of the tumor as they are highly chemosensitive. Granuloma could also be identified during intra-operative diagnosis in our study and anti-tubercular drug was administered avoiding gross total resection.
The diagnosis of PXA, a common neoplasm in older children and adolescents, can be difficult both on squash and frozen due to marked cellular pleomorphism. The spindle cell architecture and eosinophilic granular bodies are better appreciated on FS rather than squash. High-grade ependymomas are often difficult to identify by intra-operative diagnostic modalities. A case of anaplastic ependymoma was misdiagnosed as medulloblastoma on squash but correctly identified on frozen due to presence of perivascular architecture of cells forming pseudo-rosettes. A case of myxopapillary ependymoma was misdiagnosed as grade II ependymoma on squash because of scanty myxoid matrix. It was however correctly diagnosed on FS. A case of Grade II ependymoma was diagnosed as high grade glioma. Posterior fossa neoplasms are extremely common in the pediatric age group. Performing only squash cytology in these cases may lead to misinterpretation of the granular cell as PNET, lymphomas, SRCT, or embryonal cell tumors. Here, complementing squash cytology with FS helps to reduce inaccurate diagnosis. Clinical and radiological correlation becomes important in these cases.
Craniopharyngioma was unambiguously diagnosed both on squash and frozen. It was difficult to smear and had machine oil appearance in some cases. Central neurocytoma was missed both on FS and squash and was misinterpreted as low grade glioma.
Skull-based tumors are commonly encountered in pediatric population and these are deep seated and difficult to dissect. Chordoma, a rare tumor of pediatric age group could be diagnosed on smears and FS due to presence of characteristic physalipharous cells. Radiological findings further supported the diagnosis. Another skull-based tumor found in our study showed features of SRCT, however a final diagnosis of rhabdomyosarcoma could be only given on histopathological examination. ERMS are immensely chemosensitive and has a favorable outcome. A case of rhabdoid meningioma was misdiagnosed as atypical meningioma both on FS and squash due to very scanty rhabdoid cells. Final diagnosis was only possible on histology. Mesenchymal tumors, peripheral nerve sheath tumours, fibroblastic meningioma, hemangioblastoma, and hemangiopericytoma are difficult to spread on squash preparation. Arteriovenous malformations and vascular tumours like cavernomas may only show blood elements on squash. In these cases, FS is the preferred mode for IOD.,
In ganglioglioma, ganglion cells are difficult to be identify both on squash and FS leading to cell type error. In our series, a case of ganglioglioma was diagnosed as low grade glioma (WHO grade I) as the ganglion cells could not be appreciated on the smear and FS. A temporal lobe lesion showed presence of oligodendrocyte like glial cells and floating neurons on histopathology, which was a case of dysembyoplastic neuroepithelial tumor. It was misinterpreted as oligodendroglioma both on squash and FS.
Sometimes, pilocytic astrocytoma, a common tumor in children is difficult to distinguish from grade II astrocytoma on squash alone. FS is more helpful in these cases. This is important as with an accurate diagnosis, a subtotal/gross total resection can be planned per-operatively. Sometimes, distinguishing normal brain parenchyma from low grade glioma can be extremely difficult even on FFPE, so for IOD, a FS is useful to steer us away from incorrect diagnosis. In cases of metastatic deposits in the CNS, mainly seen in adults, FS is more useful as it is not possible to assess the architectural features and tumor–brain interface. Cellularity, extra-cellular environment, tumour cell infiltration, and the architecture are best appreciated on FS.
In our study, we observed that when availability of tissue is limited, squash is perhaps better for astrocytic tumors, whereas for ependymomas and non-glial tumors like germinoma, hemangiopericytoma, and rhabdomyosarcoma, FS shows better results.
The spectrum of CNS lesions in pediatric population is more varied than that of adults. From our study, we conclude that although FS and squash cytology show similar specificity and PPV, FS has overall higher sensitivity and NPV than squash in diagnosis of pediatric CNS tumors. However, a definitive intra-operative diagnosis should be made after correlating the IOD findings with the clinicoradiological data.
Financial support and sponsorship
Funded by West Bengal, DBT Project.
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]