Diagnostic Performance Comparison of Liquid-Based Preparation Methods in Thyroid FNAs

Abstract

Introduction:

Fine needle aspiration (FNA) is recognized worldwide as the primary diagnostic method for evaluating thyroid nodules. Samples collected by FNA can be spread directly onto slides, prepared for cell blocks, or processed as liquid-based cytology. Advocates of the traditional smear technique emphasize that background material such as colloid, cell sequencing characteristics in cell clusters, and cellularity is important in cytological diagnosis. They state that these properties are not observed in liquid-based liquids. Liquid-based cytology techniques aim to provide standardized slides of homogeneous cellular smears with well-preserved cell morphology, producing more precise visualization, shorter interpretation time, and more reproducible results. This study aimed to investigate and compare the diagnostic performance of liquid-based thyroid FNA samples prepared with SurePath and Cytospin over the last 6 years.

Methods:

Patients who underwent ultrasound-guided FNA between January 2015 and December 2021 were included in the study. Cytology reports, pathology reports, and demographic information of the patients were collected from the Hospital Information Management System. The programs “SPSS for Windows 21” and “MedCalc Diagnostic Test Evaluation Calculator” were used for data analysis. Chi-square tests were used in crosstab analysis. The Kolmogorov-Smirnov two-sample test, two proportions test, and the confidence interval tests were used to analyze the variables of the methods. P < .05 was considered statistically significant.

Results:

The study population comprised 4,855 patients, 83.8% female and 16.2% male. There were no statistical differences in age and gender distribution in the population of both methods. Nondiagnostic outcome rates were 11.2% for SurePath and 12.8% for Cytospin. Sensitivity, specificity, and accuracy values for SurePath and Cytospin methods were determined as 58.57%, 98.28%, and 91.12% and 79.07%, 96.88%, and 94.03%, respectively.

Conclusion:

In our study, although the qualitative analytical results were slightly in favor of the cytospin method, we think both ways can be used safely, especially in hospitals where the clinics performing sampling activities are variable and many, as well as in pathology laboratories where the workload is intense.

INTRODUCTION

Fine needle aspiration (FNA) is accepted worldwide as the primary diagnostic method for evaluating thyroid nodules due to its simplicity, reliability, precision, noninvasiveness, and cost-effectiveness.[1-5] The technique was first applied at the Memorial Hospital in New York in the 1930s. It was redefined and developed as an invaluable diagnostic tool in Sweden in the 1950s and 1960s.[6] Since then, this method has spread worldwide as an outpatient procedure. Samples collected by FNA can be applied directly onto slides, cell block prepared, or processed as liquid-based cytology.

Liquid-based cytology techniques consist of collecting aspirates in a specially developed liquid fixative; subsequent removal of cell debris, red blood cells, and inflammatory cells; homogenization; and finally, a sampling and slide production step by applying vacuum or by a settling method. It provides standardized slides of homogeneous cellular smears with well-preserved cell morphology, clearer visualization, shorter interpretation time, and more reproducible results between pathologists.[2]

Although many researchers now agree that applying Liquid-based preparation (LBP) to thyroid FNAs is acceptable, it is still being determined to what extent we can rely on the results of LBP, whether it is appropriate to use LBP alone or whether LBP should be administered in combination with Conventional cytology (CS). Although there are studies on thin-preparation and SurePath preparation of thyroid FNAs in the literature, there are few studies on Cytospin preparation. A study comparing SurePath and Cytospin techniques in thyroid FNAs could not be determined.

This study was planned in a regional hospital where an average of 1.4 million patients were examined annually, and thyroid FNA was performed in amounts ranging from 600 to 1,100. SurePath and Cytospin methods were used in the last 6 years to prepare thyroid FNAs due to changes in hospital service conditions. In this study, our objective was to evaluate the diagnostic performance of liquid-based thyroid FNA samples prepared with these two different methods and to compare the methods.

MATERIALS AND METHODS

Patients who underwent ultrasound-guided FNA in the interventional radiology unit between January 2015 and December 2021 were included in the study. FNA results from reports, pathology reports, and demographic information of the patients were collected via the Hospital Information Management System.

The study used pathological results from thyroidectomy surgery samples as a reference value. Borderline thyroid tumors (NIFTP and WDT-UMP) were considered and separately evaluated as malignant and benign.

Processing of FNA samples with Cytospin and SurePath methods:

In the SurePath (TriPath Imaging, Burlington, NC, USA) method, FNA samples were collected in an ethanol-based solution (CytoRich). The tube was incubated at room temperature for 30 minutes. It was then centrifuged at 600 g for 10 minutes. The supernatant was discarded and the material was vortexed for 6 seconds and resuspended in 6 mL of distilled water, then centrifuged at 600 g for 5 minutes. The supernatant was then discarded again and the resulting pellet was vortexed for 6 seconds, then processed on a BD PrepStain slide processor and stained with PapStain.

The Cytospin method collected FNA samples in a tube containing CytoRich Red solution. Samples were loaded onto an automated cytospin machine (Shandon cytospin 4, Thermo Electron Corporation, Pittsburgh, USA) following the manufacturer’s instructions and centrifuged for 5 minutes at 600-800 rpm. Slides prepared with the cytospin technique were fixed in 95% ethanol for 20-30 minutes and stained with PapStain.

The following performance characteristics were evaluated to demonstrate the quality of qualitative analytical results:[7] sensitivity (SS), specificity (SP), positive predictive value (PPV), negative predictive value (NPV), accuracy, Youden index (Y), the odds ratio of positive results (LR+), and the odds ratio of adverse outcomes (LR-).

Criteria applied in performance analysis and statistical interpretation:

1. Cases reported as Bethesda-V (according to the recommendations of the American Thyroid Association) were included in the malignant category for statistical analysis, as they led to surgical treatment.

2. Bethesda I, III, and IV cases were analyzed only for reporting and malignancy rates.

3. Invasive and occult tumors detected at non-FNA localization sites were excluded.

4. The data analysis used the SPSS for Windows 21 program and MedCalc Diagnostic test evaluation calculator programs.

5. Shapiro-Wilk’s test was used to determine the conformity of the variables to the normal distribution.

6. An independent samples t-test was used to compare the mean ages according to the methods.

7. Chi-square tests were used in the analysis of the crosstabs created.

8. A two-sample Kolmogorov-Smirnov test was performed to analyze the variables related to the methods.

9. Two ratio tests (two proportions test and confidence interval) were performed to compare the ratios of the diagnostic categories between the methods.

10. The number (%) for qualitative data and mean ± standard deviation statistics for quantitative data were used to summarize the data.

11. P < .05 was considered statistically significant.

RESULTS

The total study population consisted of 4,855 patients, 4,069 (83.8%) female and 786 (16.2%) male. The age range of the patients was between 9 and 76 years, and the female/male ratio was 5.1:1. The highest number of cases were seen in the 41-60 years age group (52.5%), followed by the 61 years and more (24.7%) age group. The age and sex distribution of all the cases studied with both methods are shown in Table 1.

Table 1.

Age and sex distribution table of the cases that make up the methods

Method				Old				Total
				0-18	19-40	41-60	61
SurePath	SEX	MAN	Count	8	54	229	162	453
			% of Total	0.3%	1.9%	7.9%	5.6%	15.6%
		WOMAN	Count	30	565	1,268	582	2,445
			% of Total	1.0%	19.5%	43.8%	20.1%	84.4%
	Total		Count	38	619	1,497	744	2,898
			% of Total	1.3%	21.4%	51.7%	25.7%	100.0%
Cytospin	SEX	MAN	Count	3	40	192	98	333
			% of Total	0.2%	2.0%	9.8%	5.0%	17.0%
		WOMAN	Count	18	389	860	357	1,624
			% of Total	0.9%	19.9%	43.9%	18.2%	83.0%
	Total		Count	21	429	1,052	455	1,957
			% of Total	1.1%	21.9%	53.8%	23.2%	100.0%
Total	SEX	MAN	Count	11	94	421	260	786
			% of Total	0.2%	1.9%	8.7%	5.4%	16.2%
		WOMAN	Count	48	954	2,128	939	4,069
			% of Total	1.0%	19.6%	43.8%	19.3%	83.8%
	Total		Count	59	1,048	2,549	1,199	4,855
			% of Total	1.2%	21.6%	52.5%	24.7%	100.0%

In cases prepared with the SurePath method, among the 2,898 FNA samples, 324 (11.2%) were not diagnostic, 2,241 (77.3%) benign, 212 (7.3%) atypia of uncertain significance (AUS/FLUS), 25 (0.9%) suspicious of follicular neoplasm, 64 (2.2%) suspected malignant, and 32 (1.1%) malignant. In cases prepared by the cytospin method, of 1,957 FNA samples, 250 (12.8%) were nondiagnostic, 1,444 (73.8%) benign, 187 (9.6%) AUS/FLUS, 18 (0.9%) suspicious of follicular neoplasm, 35 (1.8%) suspected to be malignant, and 23 (1.2%) classified as malignant neoplasms. These distributions are shown in Table 2 and Figure 1.

Table 2.

Distribution table of cytological diagnoses according to methods

			Cytological diagnosis						Total
			Nondiagnostic	Bethesda 2	Bethesda 3	Bethesda 4	Bethesda 5	Bethesda 6
Method	SurePath	Count	324	2,241	212	25	64	32	2,898
		% within Method	11.2%	77.3%	7.3%	0.9%	2.2%	1.1%	100.0%
	Cytopsin	Count	250	1,444	187	18	35	23	1,957
		% within Method	12.8%	73.8%	9.6%	0.9%	1.8%	1.2%	100.0%
Total		Count	574	3,685	399	43	99	55	4,855
		11.8%	75.9%	8.2%	0.9%	2.0%	1.1%	100.0%
		% within Method	P=0.094	P=0.005	P=0.006	P=0.876	P=0.352	P=0.890
		X2 P=12.682 s.d=5 P=0.027

Figure 1.

Distribution chart of cytological diagnoses according to methods

Histopathological diagnoses of 507 patients (SurePath: 314 and Cytospin: 193) who were operated on after FNA were obtained. Of the cases with the final histopathological diagnosis, the number of benign lesions was 310, the number of malignant lesions was 168, and 29 patients were in the borderline category. The distribution of the cytological and histopathological diagnosis is shown in Table 3.

Table 3.

Distribution table of the cytological and pathological diagnoses of the cases according to the methods

Count
Method			Cytological diagnosis						Total
			Nondiagnostic	Bethesda 2	Bethesda 3	Bethesda 4	Bethesda 5	Bethesda 6
SurePath	Pathological diagnosis	Hyperplasia/goitre	10	146	24	1	1	0	182
		Thyroiditis	1	4	6	0	2	0	13
		Benign thyroid neoplasms	0	5	2	1	0	0	8
		Borderline thyroid tumors	0	13	2	0	3	0	18
		Occult neoplasms	3	18	7	1	6	0	35
		Papillary thyroid carcinoma and variants	0	14	6	4	15	17	56
		Follicular carcinoma	0	0	0	1	0	0	1
		Medullary carcinoma	0	0	1	0	0	0	1
	Total		14	200	48	8	27	17	314
Cytospin	Pathological diagnosis	Hyperplasia/goitre	7	63	9	2	1	0	82
		Thyroiditis	0	5	2	3	2	0	12
		Benign thyroid neoplasms	3	5	4	1	0	0	13
		Borderline thyroid tumors	0	6	3	0	2	0	11
		Occult neoplasms	3	19	6	0	1	1	30
		Papillary thyroid carcinoma and variants	2	4	5	3	14	15	43
		Follicular carcinoma	0	0	0	1	0	0	1
		Oncocytic (Hürthle cell) tumors	0	0	0	0	1	0	1
	Total		15	102	29	10	21	16	193
	SurePath + Cytospin, total		29	302	77	18	48	33	507

The state tables describing the performance of the analytical method for the analytical scope are shown in Table 4.

Table 4.

Concordance results of the cytological result with the histopathological diagnosis

Preparation methods	Diagnostic concordance status	Cytological diagnoses (Bethesda 2, Bethesda V, and Bethesda VI)
			Positive (pc)	Negative (nc)	Total
SurePath	Concordance according to the histopathological results (when borderline tumors are considered benign)	Positive (P)	tp=38	fp=6	P=44
		Negative (n)	fn=16	tn=184	n=200
		Total	pc=54	nc=190
		Positive (P)	tp=33	fp=5	P=38
Cytospin		Negative (n)	fn=3	tn=97	n=100
		Total	pc=36	nc=102
SurePath	Concordance according to the histopathological results (when borderline tumors are considered malign)
		Positive (P)	tp=41	fp=3	P=44
		Negative (n)	fn=29	tn=171	n=200
		Total	pc=70	nc=174
Cytospin		Positive (P)	tp=35	fp=3	P=38
		Negative (n)	fn=9	tn=91	n=100
		Total	pc=44	nc=94

Sensitivity, specificity, PPV, NPV, and accuracy values for SurePath and Cytospin methods were determined as 58.57%, 98.28%, 86.61%, 92.5%, 91.12% and 79.07%, 96.88%, 2.82%, 96.05%, 94.03%, respectively. The analytical performance results of both methods are shown in Table 5.

Table 5.

The qualitative analytical performance results in the methods are as follows

Statistic (P: 20%)	Borderline Ts is malign Value - 95% CI			Borderline Ts is benign Value - 95% CI
	SurePath	Cytospin	LBC Com.	SurePath	Cytospin	LBC Com.
Sensitivity (SS)	58.57%	79.07%	66.37%	70.37%	91.43%	78.65%
	46.17% to	63.96% to	56.88% to	56.39% to	76.94% to	68.69% to
	70.23%	89.96%	74.99%	82.02%	98.20%	86.63%
Specificity (SP)	98.28%	96.88%	97.78%	96.84%	95.19%	96.26%
	95.04% to	91.14% to	95.23% to	93.25% to	89.14% to	93.40% to
	99.64%	99.35%	99.18%	98.83%	98.42%	98.12%
Positive Predictive Value (PPV)	86.61%	82.82%	85.05%	80.93%	78.37%	80.02%
	67.44% to	61.02% to	71.84% to	65.46% to	60.49% to	68.94% to
	95.29%	93.69%	92.69%	90.48%	89.55%	87.84%
Negative Predictive Value (NPV)	92.57%	96.05%	93.85%	94.49%	98.31%	95.95%
	90.40% to	93.14% to	92.17% to	91.91% to	95.18% to	94.07% to
	94.27%	97.75%	95.19%	96.28%	99.42%	97.24%
Accuracy (E)	91.92%	94.03%	92.75%	92.61%	94.59%	93.44%
	87.77% to	88.70% to	89.68% to	88.57% to	89.42% to	90.48% to
	95.02%	97.34%	95.14%	95.56%	97.71%	95.71%
Positive Likelihood Ratio (LR+)	33.97	25.30	29.87	22.28	19.02	21.02
	10.88 to 106.11	8.22 to 77.89	13.39 to 66.61	9.95 to 49.90	8.04 to 44.99	11.65 to 37.92
Negative Likelihood Ratio (LR-)	0.42	0.22	0.34	0.31	0.09	0.22
	0.32 to 0.56	0.12 to 0.39	0.27 to 0.45	0.20 to 0.46	0.03 to 0.27	0.15 to 0.33
Youden index	0.56	0.75	0.63	0.66	0.86	0.71

There was no difference between the age and sex distributions of the subjects in the methods [Pearson Chi-Square, 0.213 (<0.050), 0.217 (<0.050)].

The results of the two-sample nonparametric Kolmogorov-Smirnov test were determined as 0.763 (>0.050) to 0.626 (>0.050) and 0.662 (>0.050) in terms of the distribution of cytological diagnosis and the confirmation rates of cytological diagnosis in both groups, respectively.

The concordance between the cytological results and the pathological diagnosis of both methods is shown in Table 4. The results of the two proportions test and the confidence interval test applied in the cross-tables analysis were determined as 0.171 (>0.050) and 0.212 (>0.50).

DISCUSSION

Thyroid nodules are a common clinical problem. Epidemiological studies indicate that the prevalence of palpable thyroid nodules is approximately 5% in women and 1% in men who live in regions with sufficient iodine worldwide.[8] In 1988, Urgancıoglu et al.[9] reported the average prevalence of thyroid nodules in Turkey at 1.8% (1.4%-5.2%). Thyroid cancer accounts for 6% of cancers in women and less than 3% in men in the United States.[1] According to data from Turkey’s 2019 cancer statistics, it represents 11.9% of female and 2% of male cancers. The malignancy rate in thyroid nodules is between 1.6% and 16.0.[1,5,10,11] Although it has a good prognosis, 10%-15% of patients have a recurrence, and approximately 5% of patients who do not respond to radioactive iodine have the potential to metastasize to regional lymph nodes and distant or even distant sites.[1] For this reason, patients need appropriate triage for early detection of thyroid cancers and thyroid nodules.

Liquid-based cytology techniques have been used with increasing frequency since the FDA approved the use of ThinPrep in nongynecological specimens in 1991.[3,12,13] Liquid-based cytology or thin-layer cytology, ThinPrep (Hologic, Marlborough, MA, USA), SurePath (BD Diagnostics-TriPath Imaging, Burlington, NC, USA), and Shandon Cytospin preparation methods (Thermo Electron Corporation) aim to produce standard slides. All three techniques consist of collecting aspirates in a specially developed liquid fixative; subsequent removal of cell debris, red blood cells, and inflammatory cells; homogenization; and finally, a sampling and slide production step, either by applying a vacuum or by a settling method. It provides standardized slides of homogeneous cellular smears with well-preserved cell morphology, clearer visualization, shorter interpretation time, and more reproducible results between pathologists.[2]

The results of FNA are reported in a standard way using the Bethesda System for Reporting Thyroid Cytopathology (BSRTC) criteria and provide information on the rate of malignancy, which is crucial to guide subsequent treatment. Each category has a risk of malignancy and a usual management scheme, although actual management will depend on factors other than fine-needle sampling results. The BSRTC aims to identify patients who require surgical excision of thyroid lesions from patients who can be treated conservatively.[14,15]

What is expected from a procedure is that it allows early detection of malignancies and triage of patients for appropriate management of thyroid nodules. The procedure’s success depends on distinguishing the presence or absence of neoplastic growth in the samples. Therefore, to detect false negative or false positive results when analyzing test results, the particular categories (B-II and B-VI) in the Bethesda classification and the cases reported as ‘B-V’ in cytology were included in the statistical analysis.[8,15] The categories (B-I, B-III, B-IV) where the test result did not provide sufficient confidence and more study was required were also analyzed in the uncertainty region category.

In the meta-analysis conducted by Yosef Chong et al. in 2017,[2] they stated that sensitivity was reported between 0.49 and 1.00 and specificity between 0.58 and 1.00 in liquid-based cytology results studied with both TP and SurePath methods. Based on the results of this study, they reported that it was superior to conventional methods in terms of sample adequacy, and the sensitivity and specificity of LBC were similar to or slightly superior to those of CS. They did not significantly affect the precision of diagnosis.

In the study conducted by C. Geers et al. in 2011,[13] the sensitivity of the LBC (SP) results was 77%, the specificity was 81%, the precision was 80%, PPV 51%, NPV 93%, and it reported that the SP method was reliable in TIAS evaluations. In the study by E. Machala et al. in 2018,[4] the sensitivity and specificity were reported as 60.28% and 98.05%, PPV 90.1%, and NPV 89.35%, respectively. As a result, FNA was a suitable method to differentiate a malignant lesion from others, with an accuracy of 89.46% reported. However, in Britt-Marie Ljung’s study, some of the studies conducted with Liquid-based cytology (LBC) techniques indicated lower definitive diagnosis rates compared to conventional methods and emphasized the need for larger-scale, comparable, controlled studies.[12]

In their study in 2020, Zhao et al.[7] stated that the LBC sedimentation preparation technique is better because it can increase the PTC detection rate, and the cytological result based on this technique is well compatible with the postoperative histopathological diagnosis. They also reported that the sensitivity of liquid-based smears is higher in the preoperative diagnosis of PTC (close to 100% for categories of TBSV and TBSVI). Additionally, both LBP techniques can demonstrate PTC nucleus morphology and structural characteristics, in which the papillary tumor is flat on membrane-based smears and presented as a three-dimensional structure that is easy to identify on sedimentation smears. Therefore, they emphasized that a better understanding of subtle differences in the assessment of cell morphology and structure using membrane-based and sedimentation LBP techniques aids in the definitive diagnosis of PTC.

Yanli Zhui et al. (2020),[5] for ‘NIFTP = Ca’, when DC V and DC VI both show positive findings, sensitivity, specificity, PPV, NPV, and diagnostic accuracy are 98.3, 30.9%, 94.9%, 58.3%, and 93.5%, respectively. As NIFTP = Ca, they reported 98.0%, 84.0%, 99.4%, 58.3%, and 97.5%, respectively. They noted that NIFTP did not significantly affect diagnostic accuracy. Uresin et al.[16] reported sensitivity as 57.9%, specificity as 88.10%, PPV 52.38%, NPV 90.4%, and accuracy as 82.52% in the FNA diagnostic accuracy study performed. Researchers stated that there was a moderate agreement between cytological and pathology results, highlighting the need for more sensitive approaches.

The ability of the test to show actual patients in a group known to be sick is called the “sensitivity” of the test, and the power of the test to show actual patients among individuals known to be healthy is called the “specificity” of the test. Accuracy: The ability of a test to accurately distinguish between sick and healthy cases. In our study, the SS, SP, and E values were determined as 58.57%, 98.28%, and 91.12% for the SurePath method and 79.07%, 96.88%, and 94.03% for the Cytospin method, respectively. All detailed values are monitored in Table 5. Although the sensitivity of the SurePath preparation method was slightly low, there was no significant difference between the two methods regarding specificity and diagnostic accuracy.

Test Result Likelihood Ratio (Likelihood ratio): It is used as a value that increases the precision to obtain the cases. It is a predictive technique. Positive test result odds ratio (LR+): The ratio of the test to the illusion of reporting the disease when it says it exists (the accuracy rate of diagnosing the disease). The higher this ratio, the better differentiated actual patients. Negative test result odds ratio (LR-): It is the accuracy rate of a healthy diagnosis. The smaller this ratio, the better it can distinguish true health. Our study determined the LR+ and LP- values at 33.97 and 0.42 for the SurePath method and 25.30 and 0.22 for the Cytospin method, respectively. Both methods produce similar and successful results in the differentiation of real robust patients.

The indicator reflecting the power to reflect the reality of the result of a test is considered the predictive power (or predictive value). Sensitivity and specificity determine the test’s power to discriminate between actual patients and healthy individuals, while predictive power measures how well what the test says is true or healthy. In short, the positive predictive value is the question “What is the probability that this patient is a real patient?” that answers the question. The negative predictive value is, in the same way, when faced with a negative test result, “What is the probability that this person is healthy?” forms the answer to the question. In our study, PPV was 86.61% and NPV was 92.57% in the SurePath method. In the Cytospin method, PPV was 82.82% and NPV was 96.05%. There was no statistical difference between both methods. The predictive power of both methods was reasonable and agreed with studies in the literature.

Another benchmarking method is the Yuden INDEX (J). The J value ranges from -1 to +1. It will indicate that the test has no diagnostic power if it is less than zero. When comparing the strengths of two separate diagnostic tests, it can be said that the diagnostic test with a higher J value is roughly more effective. In our study, J = 0.56 for the SurePath method and J = 0.75 for the Cytospin method. As a result, both tests’ diagnostic power can be considered sufficient, and the Cytospin method is slightly more effective than SurePath.

Since most thyroid nodules in the nondiagnostic outcome category are not resected, it is difficult to calculate the risk of malignancy in these cases. The general malignancy rate in undiagnosed thyroid is between 1.7% and 33.4% in FNAC in the literature.[2,11,17] The rate of nondiagnostic outcomes in SBS (TP) studies (seven studies; 14,251 cases) (24.0%) is higher than the traditional method study (33.4%) reported significantly lower. Similarly, in CP studies (two studies, 730 cases), the rate of nondiagnostic outcomes in CP (7.1%) was reported to be significantly lower (13.2%) than in CS (P < .02).[2] Surgically resected nodules with ND/UNSa results are a selected subgroup with recurrent biopsies and or clinical/sonographically worrisome features. Therefore, the rate of surgically resected nondiagnostic nodules is considered to have a higher malignancy rate than the entire ND cohort. A reasonable malignancy rate in this category is reported to be 5%-10%.[15]

Our study reports nondiagnostic outcome rates as 11.2% in the SurePath method and 12.8% in the Cytospin method. There were no statistical differences between the methods, P = .094 (>.050). When incidentally detected occult carcinomas are excluded, the rates of malignant cases in these groups are 0% and 13.3%, respectively.

One of the main challenges in thyroid cytology is the indeterminate Bethesda 3 and 4. It is called the grey zone of thyroid FNA lesions. In the literature, 20%-25% of thyroid nodules belong to uncertain categories.[18] In the 2017 BSRTC, malignancy risks by diagnostic category are 6%-18% and 10%-40%, respectively, with NIFTP (=Ca) status in categories B-III and IV. In the case of NIFTP (=Ca), it is specified as 10%-30% and 25%-40%.[15] In their study, Yin Ren et al. (2020)[19] reported the malignancy rates for the categories B-III and B-IV as 15% and 21%, respectively. Although the Bethesda Thyroid Cytopathology Reporting System provides standardization in some categories, the AUS/FLUS category continues to include subjective cytological criteria. And subsequent malignancy rates in different reports are highly variable. In our study, the diagnosis rates in the uncertain category were 8.2% in the SurePath method and 10.5% in the Cytospin method when incidentally detected microcarcinomas were excluded. Malignancy rates in these categories; It was determined as 14.5% for B3 and 62.5% for B4 using the SurePath method and 17.2% and 40.0% for the cytospin method, respectively. Compared to the literature, the rate of cases in the uncertain category is slightly lower; however, the high rate of malignant cases among these cases has once again demonstrated the necessity of minimizing the diagnosis rates in the uncertain category in thyroid medications.

You can see the quality indicators determined in our study and the studies examined in the literature in Table 6.

Table 6.

The quality indicators determined in the studies examined in the literature

Researchers	Method	Sensitivity	Specificity	PPV	NPV	Accuracy	ROM
							Bth-3	Bth-4
Yosef C. et al.	SurePath	49%-100%	58%-100%	---	---	---	---	---
Geers et al.	SurePath	80%	93%	55%	98%	92%	---	---
Machala E et al.	---	60.28%	98.05%	90.1%	89.35%	89.46%	---	---
Yin Ren et al.	---	---	---	---	---	---	15%	21%
Yanli Zhui et al.	---	98.3%	30.9%	94.9%	58.3%	93.5%	---	---
Uresin et al.	---	57.9%	88.10%	52.38%	90.4%	82.52%
2017 Bethesda	---	---	---	---	---	---	10%-40%	25%-40%
Kim DH et al.	SurePath	94.4%	92.3%	---	---	93.3	---	---
Lee JD et al.	SurePath	89%	98%	---	---	73.3%	---	---
Our study	SurePath	58.57	98.28	86.61	92.5	91.12	14.5	62.5
Our study	Cytospin	79.09	96.88	82.82	96.05	94.03	17.2	40.0

CONCLUSION

Workers insist on the traditional smear technique, which states that cytological diagnosis is based on soil material such as colloid, arrangement features of cells in cell clusters, and cellularity; these features are not observed in liquid-based techniques. Proponents of the liquid-based method argue that the ease and consistency of sample handling allow for faster cytological evaluation of samples and additional examinations. Although background materials such as colloids and organization in cell clusters have been noted to be lost in liquid-based techniques, our experience has shown that a small number of colloids remains, although less than conventional smears.

Although qualitative analytical results show some results in favor of the Cytospin method, we think that the diagnostic performance of both methods is at a good level. We think it can be used safely, especially in hospitals where sampling activities are variable and many and in pathology laboratories where the workload is intense.

Statement of ethics

It was approved with the decision of the Scientific Research Commission of the Eskişehir Provincial Health Directorate dated 26.01.2022.

In our study, mainly archive information obtained from hospital information management system and laboratory information management system was used. No additional examination was performed on the patient or patient samples. It is a noninterventional clinical study.

It was approved by the „T.C. ESKİŞEHİR OSMANGAZİ UNIVERSITY NON-INTERVENTIONAL CLINICAL RESEARCH ETHICS COMMITTEE” that this study was conducted ethically in accordance with the World Medical Association Declaration of Helsinki (Decision Date: 26.10.2021, Decision No: 42).

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.