Manuscript:
Introduction
Thyroid nodules represent a common problem for surgeons as well as a diagnostic challenge for pathologists. These can be detected in more than 60% of the general population, with incidence of malignancy found to be around 5%, and 3254 new cases diagnosed in the United Kingdom (UK) in 2017.1-4 The challenge facing clinicians dealing with thyroid nodules is to achieve an accurate preoperative diagnosis of malignancy, and therefore fine needle aspiration cytology (FNAC) can play an important role in the diagnostic workup. It is a relatively safe, cost effective and simple procedure. Although it is less precise than standard histological assessment, it may help avoid invasive and potentially unnecessary surgery.1
In current UK practice, FNAC specimens are reported according to the Royal College of Pathologists (RCPath) Thy grading system, first published in 2009 and revised in 2016 (Table 1) . In its latest version, the RCPath Thy system had six diagnostic categories (DCs), as it divided the neoplasm possible category (Thy3) into Thy3a (neoplasm possible-atypia/nondiagnostic) and Thy3f (neoplasm possible, suggestive of follicular
neoplasm).1 Thyroid cysts were also subcategorised within Thy1 and Thy2 grades as Thy1c and Thy2c respectively. The RCPath system is well aligned and largely comparable with the six-tiered Bethesda System for Reporting Thyroid Cytopathology (TBSRTC), and other internationally recognised reporting terminologies for thyroid FNAC (Table 1 ).5
Regardless of the terminology used, approximately 20-25% of all thyroid cytology is classified as indeterminate (Thy3a and Thy3f), in which it is not possible to differentiate between benign, malignant, or suspicious nodules.5,6,7 For some of these indeterminate lesions the current UK management guidelines recommend diagnostic removal of the affected lobe, guided by the decision of multidisciplinary team (MDT), with potential progression to completion thyroidectomy following review of the post-operative histology.8 This will inevitably lead to some benign lesions labelled as Thy3 or Thy4 being overtreated with hemithyroidectomy unnecessarily, with all the associated risks. Conversely, a high false negative rate with Thy2 cytology could lead to missed cancers and potential progression of disease. Therefore, optimising the diagnostic performance of the grading systems used in individual centres is of the utmost importance.7,8
The aim of this retrospective study was threefold. 1) to correlate pre-operative FNAC results with subsequent histology findings as the gold standard. 2) to audit the utilisation and the diagnostic performance of each category of the RCPath Thy system, including the sensitivity, specificity, accuracy and positive predictive values for malignancy within the patient population in our locality. 3) to compare our diagnostic performance results to previously published literature.
Materials and Methods

Study subjects and data Acquisition

We conducted a retrospective single-centre observational study in a UK district general hospital. As we used deidentified routinely collected data, this was not classified as research using the NHS health research authority online decision tool (accessible from www.hra-decisiontools.org.uk/research), and therefore IRB approval was not required, supplementary material, Figure . We included all patients who had thyroid resection (total or hemithyroidectomy) performed over a ten-year period, between January 2006 and December 2015. Eligible thyroid histology data was extracted from our digital pathology database and was correlated with FNAC where available. Data collected included age, gender, indication for surgery, details of FNAC, and the final histological diagnosis. If a patient had abnormal aspirate results taken from more than one nodule, the most abnormal result was used for analysis. If final histology reported incidental malignant lesions that were not sampled during the FNAC, these reports were excluded from the analysis.

Fine needle aspiration sample processing

In our institution, FNAC samples are routinely aspirated using ultrasound guidance (UGFNAC), with a minority obtained by palpation-guidance (PGFNAC), generally from the early years of the study. Each specimen contains at least an air-dried slide and an alcohol fixed slide. Samples were prepared by the conventional methods using Papanicolaou Romanowsky-type stains. The department uses a ‘Quick Dip’ Rapid Romanowsky staining kit for air-dried slides. Samples are distributed for reporting among eight general histopathologists in the department.

Data Management

All FNAC data was sorted in to six cytopathological groups according to the RCPath Thy grading system. For those cytology results from the earlier years of the study, reports were ‘translated’ into Thy categories and slides were reviewed if needed. Incidence rates of malignancy for each Thy grading and other diagnostic performance indicators were calculated and compared with figures from previously published literature and guidelines.

Statistical analysis

The performance of thyroid cytology was assessed for its sensitivity, specificity, diagnostic accuracy, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (PLR), and negative likelihood ratio (NLR). These parameters were calculated according to the statistical equations described in the supplementary material, Table A . For the purpose of yielding accurate results, patients with Thy1 results have been excluded when measuring the diagnostic performance of other categories, and were reported separately.