2018-19 Section 7-Neoplastic and Inflammatory Diseases of the Head and Neck eBook

Table 2. Meta-analysis of the rates of malignancy

formed using Stata 14 (StatCorp LLP, College Station, TX, USA). Meta-analysis of relative risks was performed with the metan com- mand using a random effects model; and of the malignancy rates with the metaprop command using a random effects model. Het- erogeneity was assessed using Higgins I 2 statistic [16] . I 2 is a mea- sure of the inconsistency between studies and describes the per- centage of variation across studies that is due to heterogeneity (dif- ferences between studies) as opposed to sampling error (random chance). The I 2 statistic ranges from 0 to 100. The relative risk was defined for each TBSRTC diagnostic category as the risk of malig- nancy using the new system (with NIFTP) relative to the risk of malignancy using the old histopathologic classification scheme (without NIFTP). The crude malignancy rates for the 4 studies (includ- ing our data set), before and after the initiation of the his- tologic category NIFTP, are depicted in Table 1. The malignancy rates obtained using the new system (with NIFTP) were significantly lower in all TBSRTC diagnos- tic categories except for the “non-diagnostic” category (Tables 2, 3). The relative risk of a diagnosis of malignan- cy (Table 2) ranged from 0.56 (benign) to 0.97 (malig- nant). The risk differences (Table 3) ranged from –0.03 (benign) to –0.28 (suspicious for malignancy). The stud- ies showed no significant heterogeneity in the evaluation of risk differences. The relative risk of malignancy showed significant heterogeneity in the “suspicious for malignan- cy” and “malignant” diagnostic categories. All 4 studies showed reductions in the ROM but showed significant variation in the extent of reduction. Table 3 shows the risk differences in rates of malignancy for all studies. Figures 1 and 2 document the Forest Plots for relative risk of ma- lignancy and risk difference for the 4 studies. The TBSRTC was developed to standardize the cyto- logic diagnosis of thyroid nodules [1] . While the TBSRTC systemhas been widely adopted in the United States, com- peting (but similar) tiered classification schemes have been developed in other countries including Italy and the United Kingdom [9] . A number of issues have been rec- ognized subsequent to the publication of TBSRTC; most were centered around the variations in reporting, risk of malignancy andmanagement of indeterminate diagnostic categories. A meta-analysis of 51 published articles com- prising 145,928 FNA specimens documented an overall malignancy rate of 27% for the atypia/follicular lesion of Results Discussion

p value RR ≠ 1

Ca- ses

RR (95% CI)

Weight Hetero- geneity I 2 ( p value)

Diagnostic category/ study

Non-diagnostic Strickland 53 0.60 (0.23–1.53)

33.3 60.0

Faquin

70 0.94 (0.53–1.68)

Layfield 21 1.00 (0.16–6.45)

6.7

Average

– 0.83 (0.53–1.36) 0.48 100.0 0.0 (0.71)

Benign Strickland 167 0.41 (0.19–0.86) Faquin 426 0.63 (0.39–1.01) Layfield 56 0.67 (0.20–2.24) FLUS Strickland 97 0.55 (0.35–0.87) Faquin 397 0.56 (0.44–0.73) Layfield 86 0.87 (0.44–1.71) Follicular neoplasm Strickland 88 0.82 (0.58–1.18) Faquin 304 0.54 (0.41–0.73) Layfield 81 0.89 (0.49–1.62) Suspicious for malignancy Strickland 94 0.52 (0.42–0.66) Faquin 179 0.72 (0.62–0.82) Layfield 29 0.79 (0.58–1.08) Malignant Strickland 156 0.94 (0.91–0.99) Faquin 450 0.97 (0.95–0.99) Layfield 39 0.87 (0.77–0.99) Average Average Average Average

32.4 58.8

8.8

– 0.56 (0.38–0.83) 0.003 100.0 0.0 (0.62)

21.5 70.1

8.5

– 0.59 (0.47–0.72) <0.001 100.0 0.0 (0.49)

25.2 63.5 11.3

– 0.70 (0.51–0.97) 0.03 100.0 52.7 (0.12)

43.6 43.6 12.8

– 0.72 (0.62–0.82) <0.001 100.0 68.5 (0.04)

25.1 68.5

6.4

A literature search was carried out to detect all prior publica- tions addressing the impact of non-invasive follicular variant of papillary thyroid carcinoma on rates of malignancy for FNA diag- nostic categories. Six studies were found on this topic [10–15] , however, only 3 reported information that can be utilized for the current meta-analysis study [10–12] . Data was extracted from these 3 studies along with the study performed by the authors and was subjected to meta-analysis. Statistical calculations were per- 0.97 (0.95–0.99) 0.004 100.0 33.2 (0.22) The table shows the relative risk of malignancy for each cate- gory, as determined by the Bethesda System for Reporting Thyroid Cytology. Relative risk is defined as the rate of malignancy using the new system (with the classification non-invasive follicular vari- ant of papillary thyroid carcinoma, NIFTP) relative to the malig- nancy rate using the old system (without NIFTP). The weights in- dicate the relative weight given to each study in forming the aver- age. FLUS, follicular lesion of undetermined significance; RR, rel- ative risk. Average

Malignancy Risk and NIFTP

Acta Cytologica 2017;61:187–193 DOI: 10.1159/000469654

130