Patients and samples
This study was approved by the Hunter New England Local Health District Human Research Ethics Committee (HREC/16/HNE/247), and all participants provided written informed consent. To collect serum, we conducted a prospective nested cohort study, enrolling patients undergoing thyroid surgery or thyroid fine-needle aspiration (FNA) biopsy for investigation or management of thyroid disease. To collect biopsy material, we conducted a prospective cohort study of patients referred for thyroid FNA biopsy at a single high-volume clinic.
In both cohorts, patients were followed after biospecimen collection to obtain a final diagnosis of their thyroid disease based on histopathology (surgical patients) or a composite clinical assessment (clinical, ultrasound and FNA biopsy) for non-surgical patients. Relevant clinical data were extracted from the medical record to correlate levels of proNGF with age, sex, presence of hyperthyroidism (defined as thyroid stimulating hormone (TSH) level <0.1 mIU/L), and thyroid histopathology.
Prior to thyroid surgery or thyroid biopsy, serum was drawn into a serum separator tube (surgery-only patients) or plain serum tube (biopsy-first patients), centrifuged to separate, then aliquoted and frozen at -80°C. Serum samples were assayed using a proNGF enzyme-linked immunosorbent assay (ELISA) (see below) on the first or second freeze-thaw cycle only. Samples were run in triplicate at 1:20 dilution (to minimize matrix effects, as recommended by the manufacturer), with positive results confirmed on a second plate; and run with an in-house quality control (QC) samples of serum spiked with recombinant human proNGF (Biosensis Pty Ltd, Adelaide, Australia). 4-parameter logistic regression curves were fit using GraphPad Prism (v7.0 California, USA). All results above the limit of detection of 0.05ng/mL (a functional limit of 1ng/mL allowing for 20x dilution) were reported as proNGF positive.
Biopsy rinse study
Consecutive consenting adults over 18 years with a thyroid nodule graded as ‘Low-’, ‘Intermediate-’ or ‘High-risk’, according to the Sonographic Pattern stratification of the 2015 American Thyroid Association , were prospectively enrolled. Each nodule was biopsied using a 25g needle with capillary action technique. After expulsion of the cellular material for diagnostic cytopathology, the needle was rinsed with 0.5 mL phosphate-buffered saline at 4°C with the addition of protease inhibitors (cOmplete Mini, Roche, Manneheim Germany, Catalogue number 046931590011, 1 tablet per 10 mL), with subsequent refrigerated centrifugation to pellet red blood cells and insoluble debris. The supernatant containing solubilised proteins was removed and stored at -80°C prior to ELISA, performed without dilution in duplicate (due to constraints on sample volume) and analysed as above. This ‘needle-rinse’ technique is established as a sensitive method of detecting the thyroid-specific proteins thyroglobulin (an established biopsy-based tumour marker for metastatic thyroid cancer) and calcitonin (an established biopsy-based tumour marker for medullary thyroid cancer), and has the advantage of preserving cytological material for diagnostic purposes whilst potentially yielding additional information from the solubilised proteins. All results above the limit of detection of 0.05 ng/mL were reported as proNGF positive.
ProNGF ELISA validation
ProNGF was quantified using an human enzyme-linked immunosorbent assay (ELISA) kit (BEK-2226; Biosensis Pty Ltd, Adelaide, Australia), with wells coated with an antibody raised against the N-terminal precursor domain of human proNGF. Heterophilic blocking antibodies were added as recommended by the manufacturer  to a final concentration of 38µg/mL.
Performance of the proNGF ELISA was confirmed using spike and recovery and linearity of dilution experiments (Additional Table 1). A mean of 96% spike recovery was obtained (range 80-128%) when assayed in the presence of supplied heterophilic blocking antibodies. Mean recovery of the in-house QC sample, which was assayed across all plates, was 98±22% for serum, and 117±20% for rinse. The between assay coefficient of variation was 20%, and the within-assay coefficient of variation (between wells) was 3.6±2.9%.
For serum, no difference in rates of proNGF detection were observed in samples collected in serum-separator (16% positive, n=95) vs plain serum tubes (12% positive, n=109) (unadjusted p=0.42; adjusted for age, sex and thyroid hormone status p=0.79), suggesting that proNGF is not sequestered in the gel layer of a serum separator tube. Additionally, no difference in levels of proNGF detection were observed in samples stored for more than 12 months (13%, n=117), compared to less than 12 months (21%, n=87) (unadjusted p=0.21, adjusted for age, sex, thyroid cancer and hyperthyroidism p=0.79), suggesting that endogenous proNGF is stable at -80°C for at least 12 months.
Power calculations were based on pilot data, using a power of 0.8 and two sided alpha of 0.05. For the serum study, to detect a 3-fold increase in proNGF levels in patients with cancer, above the background detection of proNGF cleavage prodcuts in 6-10% of healthy sera , 46 cases and 160 controls were required. For the biopsy study, the diagnostic performance of proNGF in histological speciemens generated an area under the ROC curve of 0.94 . Conservatively assuming that our tests generate an AUC ROC of 0.85, and that the minimum clinically significant value is 0.7, 28 cases with thyroid cancer and 124 benign nodules were required.
Between group comparisons were assessed categorically using the Pearson’s Chi-square test, and continuously using the Wilcoxon Rank-Sum test, with multiple logistic regression to assess for potential interaction from clinical variables. Analyses were performed using the statistical software package Stata (version 14, Statacorp, Texas, USA).
Between 2014 and 2017, 204 patients with thyroid diseases were enrolled in the serum cohort (46 cases of thyroid cancer and 158 cases of benign thyroid conditions); and between 2016 and 2018, 183 patients with 188 nodules were enrolled in the biopsy cohort (26 cases of thyroid cancer and 162 benign nodules). Demographic and clinical information regarding the two cohorts are presented in Table 1.