Reply to letter to editor concerning “tyrosine kinase inhibitors for advanced or metastatic thyroid cancer: a meta-analysis of randomized controlled trials”

To the Editor,

In a letter to the editor, Su et al.¹ mentioned several issues thought to be methodological defeats of our study entitled “Tyrosine kinase inhibitors for advanced or metastatic thyroid cancer: a meta-analysis of randomized controlled trials.”²

First, Su et al.¹ commented that tyrosine kinase inhibitors (TKIs) possess broader targets than the rearranged during transfection (RET) kinase inhibitors and, thus, are inappropriate to be discussed in medullary thyroid carcinoma (MTC) and differentiated thyroid carcinoma (DTC). Additionally, importance should be attached to treatment comparison among different thyroid cancers to achieve a convincing outcome. Our answer is yes, Su et al.¹ was correct about the inappropriate discussion of MTC and DTC as the same entity. That is why we presented the results in two subgroups.

Second, Su et al.¹ indicated that two of our included trials – Kiyota et al.³ and Schlumberger et al.⁴, used the same sample and that the included trials should be counted as five instead of six. While we reported the exact number of trials identified, we listed the two author groups in the same row in Table 1, indicating that they used the same sample. We respect Su et al.’s¹ presentation style, but it should be stressed that our presentation has reported the facts as they are.

Third, Su et al.¹ pointed out that Figures 2 and 6 in our study shows a p > 0.10 and I² < 50% indicating low heterogeneity, and therefore a fixed-effect model should be adopted rather than a random one. Using I² < 50% as the criteria for choosing a fixed effect model has been a custom-based rather than an evidence-based practice⁵. On the other hand, there is mathematical and statistical evidence to support the appropriateness of the random-effects model as a general approach for meta-analysis⁵.

Fourth, Su et al.¹ commented that our work only showed overall survival and adverse events (AEs), offering no plausible explanation for its heterogeneity. We believe Su et al.¹ suggest that the heterogeneity of each outcome should be thoroughly discussed. We welcome this criticism because our heterogeneity subsection focused on a broader view. However, an amendment of the information is not necessary in the current article because it is not the core issue or value of our work.

Finally, Su et al.¹ counted the number of AEs reported in our included trials and indicated that the most prevalent AEs should be diarrhea, hypertension, fatigue or asthenia, decreased appetite, and hand-foot skin reaction. Su et al.¹ believed selection bias might exist in our study based on their calculation. We did not use the word “prevalent” in our work as Su et al.¹ implied. For patient-centered and evidenced-based care, potentially life-threatening AEs and those that worry the patients most are to be investigated and cared for as a priority rather than simply counting numbers. Prevalence does not necessarily equal importance. We decide not to amend the information because it does not meet the purpose of preceding discussion of the current article. Evidence-based medicine aims to provide the best care for the benefit of the patients. We greatly appreciate Su et al.¹ for offering us a chance to clarify several issues.