Thyroid Nodules as a Risk Factor for Thyroid Cancer in Patients With Graves' Disease

A Systematic Review and Meta-Analysis of Observational Studies in Surgically Treated Patients

Anastasios Papanastasiou; Konstantinos Sapalidis; Dimitrios G. Goulis; Nikolaos Michalopoulos; Evangelia Mareti; Stylianos Mantalovas; Isaak Kesisoglou


Clin Endocrinol. 2019;91(4):571-577. 

In This Article

Materials and Methods

Literature Search

Observational studies were searched for in PubMed (MEDLINE), Cochrane Library (CENTRAL) and Scopus databases as well as 'grey literature' sources in English, from inception until 25 February 2019. The search for grey literature was conducted in repositories, catalogues (EThOS), websites (OpenGrey, GetNet International) and conference proceedings of major international congresses. The search strings used were combinations of the terms 'thyroid', 'nodules', 'cancer' and 'Graves' disease' in both free text and Medical Subject Headings (MeSH) format.

Study Selection

The selected studies should confirm the diagnoses of: (a) GD, (b) thyroid nodule(s) and (c) thyroid cancer. The diagnosis of GD was based on signs of hyperthyroidism, elevated serum thyroxine (T4) and triiodothyronine (T3), decreased thyroid-stimulating hormone (TSH) concentration and presence of TSH receptor antibodies (TRAb).[1] The preoperative diagnosis of thyroid nodules was determined by clinical examination and ultrasonography. Scintigraphy was not considered necessary. All patients had undergone total or near-total thyroidectomy. The postoperative diagnosis of thyroid cancer was determined by histopathology. No restrictions were set regarding the indications for thyroid surgery. Studies that did not meet the above criteria were excluded from the meta-analysis.

Data Extraction and Quality Assessment

The selected studies were reviewed by two independent researchers (AP, SM). Any discrepancies between them were resolved in agreement with a third researcher (KS). The extracted data consisted of both the number of GD patients with and without preoperatively detected thyroid nodules accompanied by the respective prevalence of thyroid cancer. A sensitivity analysis was performed excluding all the patients with malignant or suspicious cytology to diminish any confounding bias. A subgroup analysis was performed according to the number of thyroid nodules (solitary or multiple). In case the data were insufficient or not available, e-mails were sent to the authors of the original publications followed by a second one in 15-day time. The Cochrane risk assessment tool risk of bias in non-randomized studies-of interventions (ROBINS-I tool) was used to evaluate the quality of studies included in the meta-analysis.[7] Two researchers (AP and SM) independently evaluated the risk of systematic errors; any disagreement was resolved in consensus with a third researcher (KS). Finally, the assessment of publication bias was not evaluated because of the insufficient number of included studies (<10).[8]

Data Synthesis and Analysis

The statistical analysis was performed using the REVMAN 5.3. software. The random effects model was applied for the meta-analysis as high heterogeneity was expected among the studies. The results were presented as odds ratio (OR) with 95% confidence intervals (CI). The I 2 index was applied in every analysis to assess the degree of heterogeneity, considering values >50% as indicative of high heterogeneity.