Material and Method: A retrospective study was conducted to evaluate efficacy and accuracy of frozen section in head and neck cancer of 265 patients, treated at a tertiary cancer centre hospital between January 2013 to December 2014.
Results: Out of 265 cases, it was found that 12.6% of these sections showed true positivity, 6.3% false positivity, 2.9% false negativity and 78.2% true negativity. The study also shows a sensitivity of 82.05% and specificity of 96.46%.
Conclusion: Our study shows that intraoperative frozen section reports are specific and highly sensitive. We recommend a minimum of 3-4 sections, optimum cryostat temperature, good section thickness and quality staining for a good concordance rate.
During oncosurgery, it is important for the surgeon to know if the surgical margin is clear of malignant cells [2,4,5]. The report given is usually binary in nature and it just indicates the presence or absence of tumor in the sample. Based on this report, the surgeon makes a decision on whether to make additional resections or not.
The use of intraoperative frozen section evaluation of surgical margins is an accepted and frequent practice in head and neck oncosurgery [6]. Frozen section analysis is a multistep process, beginning with retrieval of the specimen from the patient by the surgeon followed by preparation of slides, their microscopic examination and possible specimen triage for further workup, and finally rendering the frozen section (FS) diagnosis. The frozen section diagnosis is usually communicated verbally to the surgeon by the pathologist via telephone, intercom, video conference [7], and sometimes in person via an intermediary (e.g., nurse or other OT personnel). Such reports enable the surgeon to decide on whether further resections are needed or not.
Cases sent to our department for frozen section evaluation included those cases for primary diagnosis as well as cases for margin status and/or lymphnode metastasis. A total of 265 patients were selected and 734 sections were analyzed. After taking tissue bits for frozen sections, the remaining tissue was used for formalin fixation and for taking permanent sections. Frozen sections are critical to immediate surgical management [6,8]. Studies have also shown that survival rates are decreased if resection margin is found to be positive [4,9]. A routine review of frozen section can improve the accuracy of the procedure [5,8].
Primary objective:
Baseline the error rates of frozen section in our
institution
Secondary objective:
Recognize factors affecting the error rates
Discuss improvements for the common factors causing
the error
Statistical analysis was performed on the collected data using the statistical software SPSS. Crosstab analysis was used to find the sensitivity and specificity values. LR+ and LR- values were derived from sensitivity and specificity. The confidence intervals were obtained using the ROC curve analysis in MedCalc software, version 18 (MedCalc Software, Acacialaan 22, Belgium, https://www.medcalc. org/). ROC curve analysis figures were cross verified using the diagnostics test evaluation tool of MedCalc.
Table I: Demographic details of cases.
Table II: Overview of the results.
The prominent sites affected were the tongue and buccal mucosa. Table III has the details of the sites. Comparison of frozen section results with the permanent sections are shown in Table IV.
Table III: Distribution of the cases.
Table IV: Comparison of frozen section results with the permanent sections.
Table V shows observed values and interval estimate for sensitivity, specificity, disease prevalence and likelihood ratios along with 95% confidence interval for these values.
There were total of 15 discordant cases. However, only one of these discordant cases had an impact on immediate clinical management. We also observed that there was no discordance in the cases where a lymph node analysis was done. A study by White and Trotter [14] has shown very low rates for discordances in case of lymph node metastases (4 out of 427 cases), which matches our experience. Similarly, a study by Hatami et al. mentions only one discordant case out of 23 lymph nodes analysed [5]. However, there are many other studies where discordance/deferral rates involving lymph node frozen sections are high [6,8].
Analyzing the reasons for the disparity, it was noted that an inadequate number of deeper sections lead to false negativity in two cases of invasive carcinoma. Many studies have also pointed out similar reasons for false negatives [6,8,10,15]. Technical pitfalls such as thickness of the frozen section and folding were the reasons that lead to false negativity in the remaining five cases. Issues in staining by rapid H & E methods (e.g. cases of normal epithelium misinterpreted as mild/moderate dysplasia) lead to false positive cases. The details of the discordant cases are described below.
Only two out of 265 cases (one carcinoma of the buccal mucosa and the other carcinoma of the tongue) showed disparity due to lack of deeper sections. These formed 0.75% of the cases. Hence, it might not be wiser to conduct an increased number of deeper section analysis and expect the overall disparity figures to reduce significantly. The time and human resource factors for doing so should also be factored in here. The researchers who have studied this particular aspect are of the opinion that the cost implications can be argued in either direction [10]. Results of other studies have also shown very low rate of disparity due to lack of deeper sections when compared with the overall cases [6,10,15]. In our opinion, considering the time and cost factor, a lower priority can be assigned to this factor.
All other false negative cases (1.88% of the cases) were either due to staining issues, section folding, thick sections and/or lack pathologists experience which contributed to misinterpretation of the sections. Studies have shown that frozen section is important and highly effective in avoiding re-operations [16,17]. In sites such as head and neck, margin clearance is crucial as recurrence can be very aggressive and difficult to treat [18]. Hence, a false negative is more costly during intraoperative frozen section analysis as it will lead to recurrences and repeat surgeries in future. So, the issues that are main causes of false negatives should be addressed with more rigour. All these issues can be rectified with proper training of the technicians and pathologists involved in the frozen section evaluation.
On the other hand, all false positives (2.6% of the total cases and 46.67% of the cases showing disparity) occurred from the issues with staining. It is important to avoid such issues from the view point of improving the accuracy of frozen sections. Caution should be exercised in maintaining a clean stain line with regular solution changes, following recommended timings, etc.
The true positive rate also deserves a discussion as this rate is 12.07%. This indicates that a considerable number of patients benefited from the use of frozen section and it underscores the importance of frozen sections in head and neck oncosurgery.
Looking at the statistical figures, the sensitivity (82.05%) and specificity (96.46%) values closely match a similar-sized study by Du E et al. [19]. Our specificity value also matches approximately with the values found in other studies [5,11,20]. As can be seen from the likelihood ratio analysis, frozen section indicates a high increase in the likelihood of the disease (LR+ > 10). Also, the LR- value is 0.19, which is above 0.1 but below 0.2. This indicates that frozen sections have been moderately effective in indicating reduction in the likelihood of the disease though the negative results cannot be considered to be conclusive [21]. The above values come close to the values found in the study by Carvalho et al. [22]. However, to avoid repeat surgery, it is desirable that the negative results be more conclusive. This aspect should be improved in our particular setup. Currently, false negatives due to technical faults contribute 71% of the total false negative cases. If we can reduce the technical errors by 50%, it can be shown that LR- figure can be improved to 0.12. This will be a very good improvement over the present condition. It should be kept in mind that samples were all intraoperative and it might not be correct to generalize the above statement beyond such cases.
Our overall accuracy rate was 94.4% (5.6% discordance). Other studies have shown the accuracy rate of head and neck frozen sections to be in the range of 96% to 98% [6,10]. Intraoperative Consultation in Surgical Pathology [23], published by Cambridge University press in 2010, has put the error rates at < 2% after consolidating results from about six relevant studies. Our accuracy rate is slightly below, but not grossly off the mark.
In conclusion, the present study shows frozen section reports are reasonably accurate and essential in the head and neck oncosurgery. The operating surgeon should also be aware of the limitations of frozen section [5,24]. We recommend a minimum of 3-4 sections, optimum cryostat temperature, good section thickness and quality staining for a good concordance rate. The analysis and discussion presented above is mainly from the perspective of a tertiary cancer center with resource constraints that has just started frozen section analysis and we believe such analysis can be useful for centers similar to ours. As observed in other studies [6,8,25], we have also experienced that periodic review of the correlation between frozen section diagnosis and permanent section diagnosis is needed for better service quality and we strongly suggest that centers, and especially those at the starting stage, follow such review process.
CONFLICT of INTEREST
The authors declare no conflict of interest. The study is not
sponsored by any institution.
1) Gal AA. The Centennial Anniversary of the Frozen Section
Technique at the Mayo Clinic. Arch Pathol Lab Med.
2005;129:1532-5.
2) Recent Concepts In Sarcoma Treatment. Ryan J, Baker L, editors.
Dordrecht: Springer Netherlands; 1988. 81-8.
3) Sparkman RS. Reliability of frozen sections in the diagnosis of
breast lesions . Ann Surg. 1962;155:924-34.
4) Haque R, Contreras R, McNicoll MP, Eckberg EC, Petitti DB.
Surgical margins and survival after head and neck cancer surgery.
BMC Ear Nose Throat Disord. 2006;6:2.
5) Hatami H, Mohsenifar Z, Alavi SN. The diagnostic accuracy of
frozen section compared to permanent section: A single center
study in Iran. Iran J Pathol. 2015;10:295-9.
6) Mahe E, Ara S, Bishara M, Kurian A, Tauqir S, Ursani N,
Vasudev P, Aziz T, Ross C, Lytwyn A. Intraoperative pathology
consultation: Error, cause and impact. Can J Surg. 2013;56:E13-8.
7) Wellnitz U, Binder B, Fritz P, Friedel G, Schwarzmann P.
Reliability of telepathology for frozen section service. Anal Cell
Pathol. 2000;21:213-22.
8) Ahmad Z, Barakzai MA, Idrees R, Bhurgri Y. Correlation of intraoperative
frozen section consultation with the final diagnosis at
a referral center in Karachi, Pakistan. Indian J Pathol Microbiol.
2008;51:469-73.
9) Cook JA, Jones AS, Phillips DE, Soler Lluch E. Implications of
tumour in resection margins following surgical treatment of
squamous cell carcinoma of the head and neck. Clin Otolaryngol
Allied Sci. 1993;18:37-41.
10) Olson SM, Hussaini M, Lewis JS Jr. Frozen section analysis
of margins for head and neck tumor resections: Reduction
of sampling errors with a third histologic level. Mod Pathol.
2011;24:665-70.
11) Esbona K, Li Z, Wilke LG. Intraoperative imprint cytology
and frozen section pathology for margin assessment in breast
conservation surgery: A systematic review. Ann Surg Oncol.
2012;19:3236-45.
12) Eldeeb H, Macmillan C, Elwell C, Hammod A. The effect of the
surgical margins on the outcome of patients with head and neck
squamous cell carcinoma: Single institution experience. Cancer
Biol Med. 2012;9:29-33.
13) David AN, Richard JZ. Interinstitutional comparison of frozen
section turnaround time. Arch Pathol Lab Med. 1997;121:559-68.
14) White VA, Trotter MJ. Intraoperative consultation/final
diagnosis correlation: Relationship to tissue type and pathologic
process. Arch Pathol Lab Med. 2008;132:29-36.
15) Walts AE, Marchevsky AM. Root cause analysis of problems in
the frozen section diagnosis of in situ, minimally invasive, and
invasive adenocarcinoma of the lung. Arch Pathol Lab Med.
2012;136:1515-21.
16) Jorns JM, Visscher D, Sabel M, Breslin T, Healy P, Daignaut
S, Myers JL, Wu AJ. Intraoperative frozen section analysis of
margins in breast conserving surgery significantly decreases
reoperative rates: One year experience at an ambulatory surgical
center. Am J Clin Pathol. 2012;138:657-69.
17) Hosseini M, Alizadeh Otaghvar HR, Tizmaghz A, Shabestanipour
G, Arvaneh S. Evaluating the accuracy of fine needle aspiration
and frozen section based on permanent histology in patients with
follicular lesions. Med J Islam Repub Iran. 2015;29:239.
18) Jaafar H. Intra-operative frozen section consultation: Concepts,
applications and limitations. Malays J Med Sci. 2006;13:4-12.
19) Du E, Ow TJ, Lo YT, Gersten A, Schiff BA, Tassler AB, Smith
RV. Refining the utility and role of frozen section in head
and neck squamous cell carcinoma resection. Laryngoscope.
2016;126:1768-75.
20) Khoo JJ. An audit of intraoperative frozen section in Johor. Med
J Malaysia. 2004;59:50-5.
21) The Patient History. 2nd ed. Henderson MC, Tierney LM,
Smetana GW, editors. New York: McGraw-Hill; 2012.30.
22) Carvalho MB, Soares JM, Rapoport A, Andrade Sobrinho J,
Fava AS, Kanda JL, Lehn CN, Walder F, Menezes MB, Negri
SL. Perioperative frozen section examination in parotid gland
tumors. Sao Paulo Med J. 1999;117:233-7.
23) Intraoperative consultation in surgical pathology. Ranchod M,
editor. Cambridge: Cambridge University Press; 2010