Analyzing agreement patterns of intraoperative central nervous system lesion reporting according to type and grade.

BACKGROUND: Neuropathology centers are expected to offer a prompt and accurate intraoperative diagnosis regarding tumor/lesion type and grade on fresh unfixed tissue. Level of diagnostic accuracy according to type and grade and also, the experience at a new center has not been reported before. AIMS: The aim of this study is to review the agreement patterns according to tumor/lesion type and grade between intraoperative and final histopathologic diagnosis in central nervous system (CNS) lesion samples received by a newly established neuropathology center at a tertiary care neuropsychiatric hospital.
MATERIALS AND METHODS: AGREEMENT BETWEEN INTRAOPERATIVE AND FINAL HISTOPATHOLOGIC DIAGNOSIS WAS CLASSIFIED AS: (I) Grade in agreement but type not in agreement; (II) grade not in agreement but type in agreement; (III) grade and type both not in agreement; (IV) grade and type both in agreement. STATISTICAL ANALYSIS: Confidence interval (CI) of agreements was calculated for various categories of neoplastic as well as non-neoplastic lesions. CI was also calculated for groups where n × p and n × (1 - p) were more than 5, i.e., fulfilled the requirement of the central limit theorem.
RESULTS: On retrospective analysis of 333 cases, 284 (85.3%) cases were categorized as neoplastic while 49 (14.7%) cases were categorized as non-neoplastic. Among the neoplastic lesions agreement was seen in 237 (83.5%) cases while 47 (16.5%) cases showed disagreement. Similarly in non-neoplastic category; 46 (93.9%) cases showed agreement while 3 (6.15%) cases showed disagreement. Of the non-neoplastic lesions, one case fell into the agreement category I, 2 in category III and 46 in IV. Among neoplastic lesions, there were 21 cases in agreement category I, 17 in II, 9 in III and 237 in IV. On analyzing the accuracy of intraoperative reporting according to tumor type, the break up was: - Astrocytic: 2 (I), 16 (II), 2 (III), 86 (IV); oligodendroglial: 8 (I), 1 (II); ependymal: 2 (III), 6 (IV); embryonal: 23 (IV); cranial and spinal nerve tumors: 2 (II), 21 (IV); choroid plexus tumors: 4 (IV); meningeal tumors: 3 (I), 1 (III), 49 (IV); metastatic tumors: 3 (I), 17 (IV); cysts (tumor-like conditions): 14 (IV); neuronal and mixed neuronal glial tumors: 1 (III); malignant lymphoma: 1 (III); sellar tumors: 17 (IV); and mixed gliomas: 5 (I).
CONCLUSIONS: This study identifies problem areas of CNS intraoperative reporting, in a new center, with reference to tumor typing and grading. It may forewarn upcoming centers of neuropathology about the potential problem areas of intraoperative reporting.

Introduction

Neurosurgical decisions are often aided by a prompt intraoperative diagnosis. Having started neuropathology diagnostic services in 2004, ours is a relatively new center. Intraoperative consultations were started in 2005. Agreement according to tumor/lesion type and grade between intraoperative and final histopathologic diagnosis in 333 neurosurgical samples received by the Department of Pathology is being reviewed here. Diagnostic accuracy of intraoperative reporting of central nervous system (CNS) lesions has been analyzed in many studies before. This study is different in that the agreement patterns at an upcoming center have been analyzed according to tumor/lesion type and grade. The common areas of errors and lessons learnt are discussed.

Materials and Methods

A retrospective analysis of agreement patterns of intraoperative CNS lesion reporting on fresh unfixed tissue with final histopathology reporting on formalin fixed tissue was performed. Records of 5 year duration, from 2005 to 2009, were accessed. Cases where both, fresh unfixed tissue for intraoperative reporting and formalin fixed tissue for final histopathology diagnosis was not available were excluded from the study. All the other cases were taken up for analysis to avoid any selection bias. Final histopathology reports (considered the gold standard used for the diagnosis of the lesions) were based solely on the findings of formalin fixed paraffin embedded tissue to achieve the masking effect. The earlier issued intraoperative report did not have a bearing on the final reports.

This retrospective study was carried out on archival material. Patient confidentiality was protected. As per the policy of the Institutional Ethics Committee, such projects are exempt from ethical review.

In all, there were 333 cases where both, intraoperative fresh unfixed and later, formalin fixed tissue samples, were received. Intraoperative reporting was done on the basis of crush smears and frozen sections both. For crush smears, toluidine blue staining was done as the first choice and if additional smears were available hematoxylin and eosin staining was also applied. Agreement between intraoperative and final histopathologic diagnosis was classified into four categories: (I) Grade in agreement, but type not in agreement; (II) grade not in agreement but type in agreement; (III) grade and type both not in agreement; (IV) grade and type both in agreement.

Pattern of analysis followed was same for neoplastic and non-neoplastic lesions. However, since there is no grade for non-neoplastic lesions, agreement for the broad categories of inflammatory/infective/malformations was taken to be an agreement of ‘grade’. If the finer details like the infection being tuberculous in nature or type of inflammation being acute or chronic also matched then it was considered as agreement of ‘type’.

For the neoplastic lesions, the agreement pattern was further analyzed for different categories of tumors: Astrocytic; oligodendroglial; ependymal; choroid plexus tumors; neuronal and mixed neuronal glial tumors; embryonal tumors; cranial and spinal nerve tumors; meningeal tumors; tumors of the hemopoietic system; tumors of the sellar region; metastatic tumors of the CNS; cysts (tumor like conditions); mixed gliomas. Intraoperative reporting of a tumor as low grade was considered to be in agreement with final grading as World Health Organization (WHO) Grade I and II and intraoperative high-grade in agreement with WHO Grade III and IV on final diagnosis.

Statistical analysis

Confidence interval (CI) of agreements was calculated for various categories of neoplastic as well as non-neoplastic lesions. CI was also calculated for groups where n × p and n × (1-p) were more than 5, i.e., fulfilled the requirement of the central limit theorem.

Results

Of the total of 333 CNS lesions received, 284 were neoplastic and 49 non-neoplastic, 289 cranial and 44 spinal, 196 intra-axial and 137 extra-axial. Maximum patients (209) belonged to 20-59 years age group, 83 were 0-19 years and the remaining (41) were more than 60 years of age.

Of the 49 non-neoplastic lesions, one case fell into the agreement category I, two cases in agreement category III and there was agreement of both ‘grade’ and type of lesion in the remaining 46 non- neoplastic lesions [Figure 1].

Figure 1

Agreement pattern of non-neoplastic lesions

Among the 284 neoplastic lesions, there were 21 cases in agreement category I, 17 in II, 9 in III and 237 in IV [Figure 2]. On analyzing the accuracy of intraoperative reporting according to tumor type, the breakup of agreement category was: Astrocytic: 2 (I), 16 (II), 2 (III), 86 (IV); oligodendroglial: 8 (I), 1 (II); ependymal: 2 (III), 6 (IV); embryonal: 23 (IV); cranial and spinal nerve tumors: 2 (II), 21 (IV); choroid plexus tumors: 4 (IV); meningeal tumors: 3 (I), 1 (III), 49 (IV); metastatic tumors: 3 (I), 17 (IV); cysts: 14 (IV); neuronal and mixed neuronal glial tumors: 1 (III); malignant lymphoma: 1 (III); sellar tumors: 17 (IV); and mixed gliomas: 5 (I) [Figure 3].

Figure 2

Agreement pattern of neoplastic lesions

Figure 3

Tumor wise agreement pattern

Combining the neoplastic and non-neoplastic lesions the distribution of agreement [Figure 4] and 95% CI was as follows: 22 (6.6%) lesions were in agreement category I with 95% CI: 3.94-9.3; category II had 17 (5.1%) cases and 95% CI: 2.74-7.5; category III with 11 (3.3%) cases showed a 95% CI: to be 1.38-5.2; 283 (85%) cases were in agreement category IV having a 95% CI: of 81.15-88.80. CI: was also calculated for groups where n × p and n × (1 − p) were more than 5, i.e., fulfilled the requirement of central limit theorem. There were only two such groups viz. Astrocytic tumors in agreement category II and IV. For agreement category II 95% CI was 8.28-21.9 and for IV it was 73.68-88.6.

Figure 4

Overall agreement pattern

On analyzing the composition of the different agreement categories it was seen that the major constituent of agreement category I were oligodendrogliomas and mixed gliomas; astrocytomas were predominant in agreement category II; agreement category III was not dominated by any particular type and was comprised of varying types of tumors viz. astrocytic, meningeal, cranial/spinal nerve, neuronal and mixed neuronal-glial, ependymal tumors and malignant lymphoma; category IV was dominated by astrocytic tumors followed by meningeal, embryonal, cranial/spinal nerve, metastases, sellar tumors, cysts (tumor-like), ependymal and choroid plexus tumors [Figure 5a–d].

Figure 5a-d

Constitution of agreement categories in neoplastic lesions

Discussion

Accuracy of intraoperative diagnosis has been studied at many reputed neuropathology centers and ranges from 83% to 95.29%.[12] This is slightly lower than the diagnostic accuracy of intraoperative reporting in general surgical specimens.[3] At our hospital, intraoperative neuropathology diagnostic services were started about 8 years back. This study was conducted with the aim of identifying problem areas of CNS intraoperative reporting in a new center like ours, with reference to typing and grading. Intraoperative pathology evaluation was done using squash smears and frozen sections both. Cytology technique has been more useful for astrocytoms, small round cell tumors and certain metastasis. Frozen section is reported to be better for the diagnosis of meningiomas, reactive lesions, ependymomas and most metastatic lesions. Using a combination of the two techniques is most beneficial.[45]

Of the non-neoplastic lesions, one case fell into the agreement category I. This was a case of tuberculous lesion on final histopathology reporting. However, no necrosis or epithelioid cell granuloma could be seen on intraoperative smears and frozen section. There were two cases in agreement category III. One of these cases was an abscess, but the area of reactive gliosis received intraoperatively was reported as a low grade glioma. The second case, an infarct, was reported as a high-grade glioma primarily because of the high cellularity and microvascular proliferation. There was agreement of both ‘grade’ and type of lesion in the remaining 46 (93.9%) non-neoplastic lesions. In non-neoplastic lesions, clues to be looked for are: The presence of macrophages should strongly suggest the possibility of a non-neoplastic process, such as demyelination, infection or infarction. An exception to the dictum that macrophages always indicate a non-neoplastic lesion is the tumor that has been treated with surgery, chemotherapy or radiotherapy, causing tissue damage to which macrophages naturally respond;[1] it is also true that macrophages may not be seen early in the evolution of an infarct. In this non-neoplastic condition capillary endothelial cells and pericytes enlarge and to some extent, multiply. The hypertrophy of microvascular cells superficially resembles the vascular hyperplasia of malignant gliomas.[6] As compared with normal parenchyma and reactive processes, low-grade astrocytomas smear more easily, showing higher cellularity, increased cell dimensions, a certain degree of pleomorphism and more evident cell borders;[7] presence of Scherer's secondary structures, cell distribution and tumor architecture in gliomas can be of help.[8]

Pattern of intraoperative reporting was analyzed for neoplastic lesions also. The number of cases received was <10 for some groups viz. oligodendroglial, ependymal, mixed gliomas, choroid plexus tumors, malignant lymphoma and neuronal-mixed neuronal glial tumors. The outcome reveals that grading and typing both were in agreement in 100% of embryonal, sellar, choroid plexus tumors and cysts/tumor like conditions, followed by meningeal (92.5%), cranial and spinal nerve (87.5%), astrocytic (81.1%) and ependymal (75%) tumors. Overall 83.5% (95% CI: 79.13-87.8) of neoplastic lesions fell in agreement category (IV) i.e., both grade and type of intraoperative reporting matched with the final diagnosis. Though grading was accurate, but precise typing could not be performed (agreement category I) in 7.4% (95% CI: 4.35-10.4) neoplastic lesions. These included all five mixed gliomas, eight oligodendrogliomas, three cases of metastasis, three meningeal and three astrocytic tumors. The mixed gliomas, oligodendrogliomas and astrocytic tumors mentioned here were all bracketed as “glial tumors”. In the authors’ experience, the trademark perinuclear haloes of oligodendrogliomas being absent on squashes or frozen sections, appreciating the prominent cytologic uniformity and roundness of the tumor cell defined in earlier studies[9] requires some practice. The three meningeal tumors in agreement category I were high-grade and it was difficult to identify the cell of origin intraoperatively. Failure to recognize atypia in meningiomas has been reported to be a potential pitfall in earlier studies also.[4] Similarly, typing of the nature of metastases was not possible in three cases. Though typing could be done accurately, grading was inaccurate (agreement category II) in 6.0% (95% CI: 3.23-8.7) neoplastic lesions. This group comprised of 16 (15.1%) astrocytic and one oligodendroglial tumor. The primary reason for inaccurate grading of astrocytic tumors was ambiguity on features of a high-grade i.e., pleomorphism, endothelial proliferation, mitosis or necrosis. The oligodendroglial tumor in this group was graded higher because of misinterpretation of proliferating vascular network. There were a total of 3.3% (95% CI: 1.13-5.2) cases where neither typing nor grading could be done intraoperatively (agreement category III). These included two astrocytic tumors and one neuronal-mixed neuronal glial tumor, which were reported as reactive gliosis; one case of rhabdoid meningioma where the rhabdoid cells were mistaken to gemistocytic and the tumor reported as low grade astrocytic tumor; two cranial/spinal nerve sheath tumors which were reported as granulation tissue; two high-grade ependymal tumors reported as low grade astrocytic; one case of malignant lymphoma as chronic inflammatory process.

Most studies do not report upon the level of correlation between intraoperative and final diagnosis. Though there have been studies assessing overall accuracy of intraoperative reporting in CNS lesions, very few have analyzed the details of it. Uematsu et al.[10] have clarified the accuracy as complete correlation (intraoperative diagnosis exactly the same as the final diagnosis); partial correlation (intraoperative diagnosis was not incorrect but was too broad to qualify as a complete correlation); no correlation (intraoperative diagnosis was incorrect and different from the final diagnosis). They concluded the accuracy of their intraoperative reporting as complete (87.8%), partial (7.8%) and no (4.4%) correlation. Our study further defines the partial correlation group as grade in agreement, but type not in agreement (I) and grade not in agreement but type in agreement (II). Moreover, both parameters i.e., type and group of the CNS neoplasm have their own clinical relevance. This also prompted the authors to analyze, in routine intraoperative reporting, which tumors are more susceptible to errors in reporting their type and which ones their grade. This awareness, it is believed, would lead to assimilation of more radiological, per operative and cytological details for bringing about an improvement in intraoperative CNS lesion reporting.

Shuangshoti et al[11] reported partial correlation in 13%, complete in 83% and none in 4%. A recent study by Ahmad et al.[12] has shown the correlation between the intraoperative frozen sections with final diagnosis on permanent sections. The concordant diagnostic frequency in their study is 97.08%. Our results are comparable to the previous studies. If agreement category IV is taken to be ‘complete correlation’ diagnostic accuracy was seen in 85% cases. Agreement categories I and II can be considered as ‘partial correlation’ and there were 6.6% and 5.1% cases respectively in these groups. ‘No correlation’ group, which is equivalent to our agreement category III had 3.3% cases. Problem areas reported in earlier studies related to accuracy of intraoperative diagnosis are: Grading of gliomas, diagnosis of undifferentiated metastatic carcinomas, primitive neuroectodermal tumors, sarcomas, rare types of histology such as gliomatosis and xanthomatous lesions,[9] failure to recognize atypia in meningiomas.[4]

While astrocytic tumors formed the major group where both grading and typing were accurate, they were also the major constituent of the category where grading did not correlate. Oligodendrogliomas and mixed gliomas were the most common tumors, which were difficult to type. Rarer variants of tumors and lesions with lymphoid cell population, spindle cell or mixed type of cellular composition were more liable to be misdiagnosed. The findings of this study may assist the upcoming centers of neuropathology as well as the post graduate students pursuing neuropathology. The impact needs to be evaluated. This study may also forewarn upcoming centers of neuropathology about the potential problem areas of intraoperative reporting in CNS lesions.

Apart from squash smears, imprint cytology is also used in few neuropathology centers, but after intensive literature search it may be concluded that both squash and imprint smear examination of lesions of CNS are useful, quick and reliable diagnostic tools. However, squash smear examination is superior to imprint smears because of its better sensitivity, specificity and diagnostic accuracy.[13] It was found that some authors have reported a better diagnostic accuracy with imprint smears[14] and their centers have acquired a remarkable proficiency in intraoperative imprint smear reporting-especially with stereotactic biopsies. However, a sizeable amount of the literature in scientific journals and textbooks also exists on squash/crush smear features.[15] In fact, some studies have also reported a superior sensitivity, specificity and diagnostic accuracy with squash smears as compared with imprint smears.[13]

To summarize, it was found that the overall accuracy of intraoperative reporting of CNS lesions at a new center does not differ significantly from the longer established ones. The problem areas encountered are almost the same. Non-neoplastic lesions provide a better intraoperative diagnostic accuracy. Among the neoplastic lesions grading of astrocytic and typing of oligodendroglial and mixed gliomas may constitute a problem area at new centeres. The lesions more likely to be misdiagnosed intraoperatively are the rarer variants of tumors or lesions having lymphoid cell population, spindle cell or mixed type of cellular composition. Appreciating the nuances of cellular and architectural morphology may better with experience. Better inputs about clinical and radiological features and experience with the appreciation of cellular features are likely to yield better outcome.

To the best of authors’ knowledge, agreement patterns of intraoperative reporting in CNS lesions have neither been analyzed in the context of newly established neuropathology centers nor qualified according to agreement for tumor grade and type. This study at a center in its initial 5 years of being established may serve as a baseline to record any temporal improvement in diagnostic accuracy of intraoperative reporting of CNS lesions.