How good are we at fine needle aspiration cytology?

A 2.5D convolutional neural network for HPV prediction in advanced oropharyngeal cancer

BACKGROUND

Infection with human papilloma virus (HPV) is one of the most relevant prognostic factors in advanced oropharyngeal cancer (OPC) treatment. In this study we aimed to assess the diagnostic accuracy of a deep learning-based method for HPV status prediction in computed tomography (CT) images of advanced OPC.

METHOD

An internal dataset and three public collections were employed (internal: n = 151, HNC1: n = 451; HNC2: n = 80; HNC3: n = 110). Internal and HNC1 datasets were used for training, whereas HNC2 and HNC3 collections were used as external test cohorts. All CT scans were resampled to a 2 mm³ resolution and a sub-volume of 72x72x72 pixels was cropped on each scan, centered around the tumor. Then, a 2.5D input of size 72x72x3 pixels was assembled by selecting the 2D slice containing the largest tumor area along the axial, sagittal and coronal planes, respectively. The convolutional neural network employed consisted of the first 5 modules of the Xception model and a small classification network. Ten-fold cross-validation was applied to evaluate training performance. At test time, soft majority voting was used to predict HPV status.

RESULTS

A final training mean [range] area under the curve (AUC) of 0.84 [0.76-0.89], accuracy of 0.76 [0.64-0.83] and F1-score of 0.74 [0.62-0.83] were achieved. AUC/accuracy/F1-score values of 0.83/0.75/0.69 and 0.88/0.79/0.68 were achieved on the HNC2 and HNC3 test sets, respectively.

CONCLUSION

Deep learning was successfully applied and validated in two external cohorts to predict HPV status in CT images of advanced OPC, proving its potential as a support tool in cancer precision medicine.

Surgeon-performed ultrasound guided fine-needle aspirate biopsy with report of learning curve; a consecutive case-series study

Background

Fine-needle aspiration biopsy has become the standard of care for the evaluation of thyroid nodules. More recently, the use of ultrasound guided fine-needle aspiration biopsy (UG-FNAB) has improved adequacy of sampling. Now there has been improved access to UG-FNAB as ultrasound technology has become more accessible. Here we review the adequacy rate and learning curve of a single surgeon starting at the adoption of UG-FNAB into surgical practice.

Methods

UG-FNABs performed at Sunnybrook Health Sciences Centre from 2010 to 2015 were reviewed retrospectively. Nodule characteristics were recorded along with cytopathology and final pathology reports. Chi-square analysis, followed by the reporting of odds ratios with confidence intervals, were used to assess the statistical significance and frequencies, respectively, of nodule characteristics amongst both diagnostic and non-diagnostic samples. A multiple regression analysis was conducted to determine if any nodule characteristic were predictive of adequacy of UG-FNABs. The learning curve was assessed by calculating the eventual adequacy rates across each year, and its statistical significance was measured using Fischer’s Exact Test.

Results

In total 423 biopsies were reviewed in 289 patients. The average nodule size was 23.05 mm. When examining if each patient eventually received a diagnostic UG-FNAB, regardless of the number attempts, adequacy was seen to increase from 70.8 % in 2010 to, 81.0 % in 2011, 90.3 % in 2012, 85.7 % in 2013, 89.7 % in 2014, and 94.3 % in 2015 (Fischer’s Exact Test, p = 0.049). Cystic (χ² = 19.70, p <0.001) nodules were found to yield higher rates of non-diagnostic samples, and their absence are predictive of obtaining an adequate biopsy as seen in a multiple regression analysis (p < 0.001) Adequacy of repeat biopsies following an initial non-diagnostic sample was 75.0 %.

Conclusions

Surgeons are capable of performing UG-FNAB with a learning curve noted to achieve standard adequacy rates. Cystic nodules are shown to yield more non-diagnostic samples in the surgeon’s office.

Surgeon performed ultrasound-guided fine-needle aspirates of the thyroid: 1067 biopsies and learning curve in a teaching center

BACKGROUND

Surgeon performed ultrasound-guided fine-needle aspirates (UG-FNAs) reduce delay in diagnosis and allow for surgeon surveillance. We present the first report on a learning curve and impact of head and neck surgical trainees on adequacy rates.

METHODS

Thyroid UG-FNA biopsies from 2009 to 2013 were reviewed retrospectively. Specimen adequacy, cytologic diagnosis, and surgical pathology were used to calculate adequacy and accuracy.

RESULTS

One thousand sixty-seven biopsies were examined in 723 individuals. The adequacy rate from adoption into practice improved from 71% to 78% to 85% over 300 cases. When UG-FNA was subsequently taught to trainees, adequacy rates varied among trainees (p < .037), and there were higher nondiagnostic rates earlier in training (p = .04). Adequacy was not related to size or palpability, but cystic lesions yielded more inadequate specimens (p < .001).

CONCLUSION

Surgeon performed UG-FNA biopsy can be performed adequately in an outpatient setting. Adequacy rates reach acceptable levels after 300 cases, whereas trainee involvement impacts adequacy rates. © 2015 Wiley Periodicals, Inc. Head Neck 38: E1281-E1284, 2016.

Routine non-thyroid head and neck cytology in a large UK centre: clinical utility and pitfalls

Objective:

This study aimed to examine the performance of head and neck cytology at Nottingham University Hospitals between 2009 and 2010.

Methods:

Cases were extracted from the Winpath pathology reporting system and correlations were investigated between results and the histological and clinical outcomes. Specimen adequacy and the sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy of the cytology tests were calculated.

Results:

In all, 19.7 per cent of aspirates were judged to be inadequate. The absolute and relative sensitivities of head and neck cytology were 87.0 per cent and 89.0 per cent, respectively, and the absolute and relative specificities were 99.0 per cent and 97.0 per cent, respectively. The positive predictive values were 99.0 per cent and 96.0 per cent and the negative predictive values were 92.0 per cent and 92.0 per cent for a diagnostic accuracy of 94.5 per cent and 93.0 per cent. The performance was consistent with previous reports and superior to that of a recent UK series. The high rate of inadequate samples is, however, a concern.

Conclusion:

Head and neck cytology is a robust technique at our institution, although there are certain problem areas. There is room for improvement in the technical quality of fine needle aspiration.

Do Immediate Cytology and Specialist Radiologists Improve the Adequacy of Ultrasound-Guided Fine-Needle Aspiration Cytology?

Objective

To assess whether a dedicated “1-stop” neck lump clinic has improved the percentage of adequate fine-needle aspiration cytology (FNAC) samples and reduced the need for repeat FNAC.

Study Design

Retrospective review.

Setting

District General Hospital in the United Kingdom.

Subjects and Methods

Patients attending for ultrasound-guided FNAC over a 6-month period from August 2012 to February 2013. Patients were placed in 4 groups: group 1, FNAC performed by any of the subspecialist radiologists with cytology support (n = 100); group 2, FNAC performed by general radiologists without cytology support (n = 112); group 3, FNAC performed by a particular subspecialist radiologist with cytology support (n = 61); and group 4, FNAC performed by the same subspecialist radiologist without cytology support (n = 125).

Results

There was a significantly higher rate of adequacy of FNAC in the presence of a subspecialist radiologist with immediate cytology (group 1) versus a general radiologist without cytology support (group 2; 87/100 vs 63/112, P = .0001), a significantly higher rate of adequacy of FNAC in the presence of cytology support with the same radiologist (group 3 vs group 4, 55/61 vs 97/125, P = .04), and a significantly higher rate of adequacy of FNAC in the presence of a subspecialist radiologist versus a general radiologist without cytology support (group 4 vs group 2, 97/125 vs 63/112, P = .0005).

Conclusion

Immediate cytology and the presence of a subspecialist radiologist increase the adequacy of FNAC. The adequacy rate of non–cytology-supported FNAC or nonsubspecialist FNAC is below the adequate rate expected from the literature or as recommended in national guidelines.

Verification bias in diagnostic accuracy studies for fine- and core needle biopsy of salivary gland lesions in otolaryngology journals: a systematic review and analysis

BACKGROUND

Diagnostic test accuracy (DTA) studies for needle biopsy are frequently published in otolaryngology journals, but this body of literature has not been assessed for verification bias.

METHODS

We conducted a systematic review of DTA studies on needle biopsy of salivary gland lesions appearing in otolaryngology journals. Studies were assessed by 2 reviewers for verification bias.

RESULTS

We identified 95 DTA studies for needle biopsy of salivary gland lesions. Eighty-one studies (84%) had verification bias. Five of the biased studies provided sufficient data to estimate the extent of bias. Verification bias was associated with an overestimate of sensitivity and an underestimate of specificity. Studies on core needle biopsy (CNB) had a lower rate of verification bias than fine-needle aspiration (FNA) studies.

CONCLUSION

Verification bias is common in DTA studies of needle biopsy for salivary gland lesions published in ear, nose, and throat (ENT) journals. Such studies overestimated sensitivity and underestimated specificity.

Neck lump clinics: is on-site assessment of fine needle aspirate diagnostic adequacy cost-effective?

OBJECTIVES

To establish the diagnostic adequacy of ultrasound-guided fine needle aspiration cytology samples at the East Berkshire neck lump clinic, and to perform a cost-benefit analysis related to the hypothetical addition of an on-site cytology technician (required to review fine needle aspiration specimen adequacy).

METHOD

The adequacy of all ultrasound-guided fine needle aspiration procedures was reviewed from 1 January to 30 June 2011. These results were used in the cost-benefit analysis related to on-site cytology assessment.

RESULTS

Of the 307 ultrasound-guided fine needle aspiration cytology procedures performed over 6 months, 67 (22 per cent) were reported to be non-diagnostic. Operator experience was found to correlate significantly with diagnostic adequacy (p < 0.001). Only 5 per cent of all fine needle aspirations were initially non-diagnostic but diagnostic on repeat sampling. This suggests that the financial and time costs of on-site fine needle aspirate adequacy assessment would outweigh any benefit.

CONCLUSION

In this series, the experience of individuals performing fine needle aspirations was the most important factor related to adequacy.

Fine needle aspiration: an atraumatic method to diagnose head and neck masses

Background

Patients presenting with a mass require tissue biopsy for histological diagnosis and treatment. Fine needle aspiration (FNA) is offered as an atraumatic, well tolerated, and inexpensive method for obtaining a biopsy from these lesions.

Objectives

In this study we evaluated the accuracy of FNA as an atraumatic method among patients with nonthyroidal masses for diagnosis of neoplastic masses compared to open surgery.

Patients and Methods

In a cross-sectional study, 65 patients with a head and neck masses (nonthyroidal) referred to us from 2004 to 2009. Those who had both FNA and open biopsy (the gold standard) were assessed for specificity, sensitivity, positive and negative predictive values of FNA in diagnoses.

Results

Sixty-five cases with both definite diagnoses of open biopsy and FNA were assessed. The mean (± standard deviation) age of patients was 39.96 ± 19.69 years (range 10 to 82 years). Twenty-five (40.8%) subjects were categorized as malignant neoplasms, 16 (19.4%) as benign neoplasms, and 24 (39.8%) as non-neoplastic lesions. The sensitivity, specificity, positive and also negative predictive values of FNA in the diagnosis of neoplastic masses were 95%, 85%, 92.68%, and 91.66% respectively, and the diagnostic accuracy was 92.3%.

Conclusions

It seems that FNA is a useful atraumatic diagnostic technique with a high diagnostic accuracy which can provide a highly sensitive diagnosis with low false positive diagnoses in patients with nonthyroidal masses.