Indian experience of AUS/FLUS diagnosis: is it different from rest of Asia and the West?-A systematic review and meta-analysis

Abstract

Background: Atypia of undetermined significance/follicular lesion of undetermined significance (AUS/FLUS) is the most heterogeneous subcategory of the Bethesda system for thyroid cytopathology with laid down permissible limits of frequency and risk of malignancy (ROM). Due to differences in thyroid clinical practice worldwide, variations have been found in frequency, resection rates (RR) and ROM. Therefore, this systematic review and meta-analysis of AUS/FLUS across different regions was conducted and comparison of data from India was done with the rest of Asia and the West.

Methods: We searched PubMed and Google search engines from January 2009-Dec 2019 using terms "AUS Thyroid" and "FLUS Thyroid". Meta-analysis was performed using DerSimonian-Laird method and 95% confidence intervals were calculated using random effects model. Independent samples t-test was used to compare frequency, RR, and ROM of AUS/FLUS between India, rest of Asia and the West.

Results: Out of 15,000 studies on internet, 60 (18 Indian, 12 Asian and 30 Western) were included. Total aspirates were 201,657; 14,279 Indian, 62,448 Asian, 124,930 Western. Pooled estimates were: frequency 7.3% (6.3-8.3%), RR 41.9% (37.4-46.6%), ROM 33.3% (26.8-39.9%). Pooled prevalence of rate did not vary significantly across the three regions. Pooled prevalence of RR was highest in India (52.9%) and lowest in rest of Asia (26.5%); of ROM was highest in Asia (45.9%), lowest in the West (26.3%) (P<0.01). Statistical analysis demonstrated publication bias, limited to Indian and Western studies. Papillary thyroid carcinoma was the most common surgical diagnosis (87.9%; 1,082/1,231).

Conclusions: This meta-analysis showed differences in thyroid clinical practice followed in India, rest of Asia and the West. Although pooled prevalence of rate of AUS/FLUS was similar across the three, pooled RR and ROM varied. AUS/FLUS nodules were more frequently resected in Indian and Western studies than in rest of Asia. ROM was higher than recommended values in all three areas, being intermediate for India.

Keywords: Atypia of undetermined significance/follicular lesion of undetermined significance thyroid (AUS/FLUS thyroid); Indian; meta-analysis; resection rate (RR); risk of malignancy (ROM).

Figure 1

Forest plots illustrating meta-analysis results of (A) frequency (I-squared =96.4%), (B) resection rate (I-squared =95.8%), (C) risk of malignancy-resection (I-squared =79.0%), and (D) risk of malignancy-overall (I-squared =70.3%) of AUS/FLUS in India. AUS/FLUS, atypia of undetermined significance/follicular lesion of undetermined significance.

Figure 2

Forest plots illustrating meta-analysis results of (A) frequency (I-squared =98.6%), (B) resection rate (I-squared =93.8%), (C) risk of malignancy-resection (I-squared =95.8%), and (D) risk of malignancy-overall (I-squared =93.5%) of AUS/FLUS in rest of Asia. AUS/FLUS, atypia of undetermined significance/follicular lesion of undetermined significance.

Figure 3

Forest plots illustrating meta-analysis results of (A) frequency (I-squared =99.1%), (B) resection rate (I-squared =97.7%), (C) risk of malignancy-resection (I-squared =93.2%), and (D) risk of malignancy-overall (I-squared =92.4%) of AUS/FLUS in West. AUS/FLUS, atypia of undetermined significance/follicular lesion of undetermined significance.

Figure 4

Funnel plots prepared for the studies included in the meta-analysis to evaluate: (A) frequency. Egger’s regression test confirmed presence of publication bias in Indian (P=0.00) and Western studies (P=0.01), but none in Asian studies (P=0.77). (B) Resection rate. Egger’s regression test showed absence of publication bias in Indian (P=0.08), Asian (P=0.38) and Western studies (P=0.35). (C) Risk of malignancy-resection. Egger’s regression test confirmed presence of publication bias only in Western studies (P=0.01), but none in Indian (P=0.26) and Asian (P=0.07) studies. (D) Risk of malignancy-overall. Egger’s regression test confirmed presence of publication bias in Indian (P=0.03) and Western (P=0.01) studies, but none in Asian (P=0.52) studies.