The Risk of Immune-Related Thyroid Dysfunction Induced by PD-1/PD-L1 Inhibitors in Cancer Patients: An Updated Systematic Review and Meta-Analysis

Background Thyroid dysfunction is common for cancer patients receiving PD-1/PD-L1 inhibitor therapies. To clarify the incidence risk of thyroid dysfunction would be important for guiding anti-PD-1 and anti-PD-L1 immunotherapy. Therefore, the updated meta-analysis was conducted to evaluate the incidence risk of thyroid dysfunction caused by PD-1/PD-L1 inhibitors. Methods PD-1/PD-L1 inhibitor related clinical trials were collected by a systematic search of the PubMed. Some relevant studies were identified by a manual search. The incidence risk of all grades and grades 3-5 was analyzed and evaluated by random effect model. The Newcastle Ottawa Scale was used for the quality assessment of all clinical trials. Results Forty-three clinical trials were collected. Compared with chemotherapy, the risk of hypothyroidism of all grades was significantly higher (OR=7.15, 95%CI:[4.85, 10.55], I2 = 40%, Z=9.91(P <0.00001)) in PD-1/PD-L1 group. Similar results could also be noted, when the control group was placebo or CTLA-4. When PD-1/PD-L1 was combined with other treatments for cancer patients, the risk of hypothyroidism of all grades was also significantly increased. Similar to the analysis results of hypothyroidism, PD-1/PD-L1 inhibitors played the same role in increasing the risk of hyperthyroidism and thyroiditis. Few significant analysis results was noted, when the risk of thyroid dysfunction of grades 3-5 was assessed. Conclusion Whether used alone or in combination with other anti-tumor drugs, PD-1/PD-L1 inhibitors increased the risk of thyroid dysfunction, especially for hypothyroidism. Furthermore, PD-1/PD-L1 was better than chemotherapy and CTLA-4 in increasing the risk of thyroid dysfunction.

Background: Thyroid dysfunction is common for cancer patients receiving PD-1/PD-L1 inhibitor therapies. To clarify the incidence risk of thyroid dysfunction would be important for guiding anti-PD-1 and anti-PD-L1 immunotherapy. Therefore, the updated metaanalysis was conducted to evaluate the incidence risk of thyroid dysfunction caused by PD-1/PD-L1 inhibitors.
Methods: PD-1/PD-L1 inhibitor related clinical trials were collected by a systematic search of the PubMed. Some relevant studies were identified by a manual search. The incidence risk of all grades and grades 3-5 was analyzed and evaluated by random effect model. The Newcastle Ottawa Scale was used for the quality assessment of all clinical trials.
Results: Forty-three clinical trials were collected. Compared with chemotherapy, the risk of hypothyroidism of all grades was significantly higher (OR=7.15, 95%CI: [4.85, 10.55], I 2 = 40%, Z=9.91(P <0.00001)) in PD-1/PD-L1 group. Similar results could also be noted, when the control group was placebo or CTLA-4. When PD-1/PD-L1 was combined with other treatments for cancer patients, the risk of hypothyroidism of all grades was also significantly increased. Similar to the analysis results of hypothyroidism, PD-1/PD-L1 inhibitors played the same role in increasing the risk of hyperthyroidism and thyroiditis. Few significant analysis results was noted, when the risk of thyroid dysfunction of grades 3-5 was assessed.

INTRODUCTION
Programmed cell death protein 1 (PD-1) and its ligand (PD-L1) inhibitors, developed to overcome the immune escape mechanisms of cancer progression and manipulate the immune system to recognize and attack cancer cells, have been widely used for cancers (1). While achieving satisfactory clinical anti-tumor treatment effects, more and more drug-induced toxic and side effects have also been reported, and more and more attention has been drawn from clinicians (1)(2)(3). Treatment guidelines for PD-1/PD-L1 related side effects have been made and used to guide clinical works (2).
Thyroid dysfunction was one of the common toxic side effects of PD-1/PD-L1 inhibitors and had been reported in plenty of clinical trials . Moreover, It was reported that the incidence of PD-1/PD-L1 induced thyroid dysfunction was related to the clinical response and the prognosis of patients (51,52). Therefore, clarifying the incidence risk of PD-1/PD-L1 related thyroid dysfunction would be of great significance for guiding clinical immunotherapy and judging the prognosis (51,52). Although thyroid dysfunction might appear in different forms (53), hyperthyroidism, hypothyroidism, and thyroiditis were still the most common manifestations (1), which were also reported most frequently in clinical trials . Due to more and more clinical trials investigating the clinical efficacy and safety of PD-1/PD-L1 in cancer patients have been finished in recent two years (4-23), we conducted this updated metaanalysis to reassess the incidence risk of PD-1/PD-L1 induced hyperthyroidism, hypothyroidism, and thyroiditis.

METHOD
The process of the meta-analysis was put into practice followed the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) (54).

Types of Enrolled Studies
Clinical trials, involving PD-1 or PD-L1 inhibitors, were identified by the PubMed search. Hematological malignancies were excluded first. Phase III clinical trials for all kinds of cancer patients would be taken as the priority. Clinical trials, reported with partial results or belonging to other phases, would be arranged in an alternative location. For all clinical trials included in the study, the control group was necessary, but there was no specific requirement for the treatment regimen of them. The results of the enrolled clinical trial must be reported in English.

Search Strategy
Just as proposed by the PRISMA, keywords (neoplasm, cancer, precancer, malignant, premalignant, tumor, PD-1, PD-L1, and clinical trial) for search were set according to the PICOS (participants, interventions, comparisons, outcomes, and study design) guidelines (54). The range of published time was set between Nov 23, 2010 and Nov 23, 2020. Four members of us were appointed for eligibility assessment and data extraction. In the case of duplicated reports of the same clinical trial, only one of them was used for the final analysis, and others would be included in the systematic review. The corresponding authors (Yuping Sun and Guohai Su) had the right to deal with all results and disagreements.

Evaluation of Study Quality and Publication Bias
Assessment for publication bias and risk of bias of individual trials were finished by Funnel plots, Egger's test, Harbord's test, and the Newcastle-Ottawa scale (NOS) (54)(55)(56)(57)(58)(59). Risk of bias summary, including selection bias, performance bias, detection bias, attrition bias, reporting bias and other bias, would be checked and shown in a single figure. A P-value of <0.05 was used as the cut-off value for statistical significance.

Outcome and Exposure of Interest
Baseline characteristics of all enrolled clinical trials, including duplicating reported ones, would be collected and summarized in a table. Grading of thyroid dysfunction, including hyperthyroidism, hypothyroidism, and thyroiditis, ranging from 1 (mild symptoms that do not interfere with activities of daily living) to 5 (fatal thyroid toxicities), was collected and gathered in excel tables. Dichotomous data would be given a priority, and other types of data would be collected first and then converted into dichotomous data.

Assessment of Heterogeneity and Statistical Analysis
Heterogeneity of all the data, identified by Cochrane's Q statistic test, was assessed by the DerSimonian-Laird method and quantified by I 2 values (54,59). Three different grades, including low, moderate, and high, were divided according to I 2 values ( < 25%, 25-50%, and > 50%). All the process of analyses was finished by the software Review Manager 5.3. The random effect model (RE) was used to deal with all the data to calculate odds ratio (OR) and their corresponding 95% confidence interval (CI) (60). The fixed effects (FE) model was only used for calculation of the funnel plots. All reported P values are 2-sided, and P<0.05 was taken to indicate statistically significance. Subgroup and stratification analyses would be performed according to tumor types, treatment regimens, and PD-1/PD-L1 inhibitors.

(Supplementary
Drug-induced thyroid dysfunction is one of the common causes of hyperthyroidism (67). Whether PD-1/PD-L1 inhibitors were used alone or in combination with other drugs, it indicated that PD-1/PD-L1 inhibitors increased the risk of hyperthyroidism of all grades ( Figures 4A-D). When PD-1/PD-L1 combined with chemotherapy was compared with PD-1/PD-L1, no statistical analysis results of hyperthyroidism of all grades was found ( Figure 4E) (11,18). Through the above analysis, we clarified the role of PD-1/PD-L1 inhibitors in increasing the risk of hyperthyroidism of all grades (Figure 4 and Supplementary Figure 4) (5-12, 14-18, 24-31, 33-35, 37-40, 42, 43, 45-50). Through subgroup analysis, high heterogeneity (I 2 = 55%) was considered to be mainly caused by PD-1 related NSCLC subgroup (I 2 = 70%, Figure 4B) (27,29). No obvious publication bias was found among all the enrolled clinical trials (Supplementary Figure 4). Though similar incidence trend could also be seen in the assessment of hypothyroidism of grades 3-5 ( Figure 5), statistical significant result was only found in ( Figure 5D). Since only two clinical trials were included ( Figure 5D), the analysis results need to be further verified.
By reviewing and analyzing PD-1/PD-L1 related literature (4-50), we found that PD-1/PD-L1 increased the risk of thyroid dysfunction. It reminds us that we need to monitor and evaluate the thyroid function status in time for patients receiving PD-1/PD-L1 treatment to prevent the occurrence of adverse events (1-3, 64-67).

Strengths and Limitations
Strengths: This meta-analysis was conducted according to the PRISMA guidelines. The literature searching process was put into practice in accordance with the PICOS principle. The quality of all enrolled clinical trials was high. Stratification and subgroup analyses were conducted as much as possible. Therefore, the conclusion was much more reliable. Limitations: First, some clinical trials related to PD-1/PD-L1 inhibitors cannot be included for meta-analysis due to obvious heterogeneity. Second, the low number of studies that reported the data of thyroid dysfunction made it difficult to get a definite conclusion.

CONCLUSION
Whether used alone or in combination with other anti-tumor drugs, PD-1/PD-L1 inhibitors increased the risk of thyroid dysfunction, especially for hypothyroidism. Furthermore, PD-1/PD-L1 was better than chemotherapy and CTLA-4 in increasing the risk of thyroid dysfunction.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
The corresponding authors (YPS and GS) had the right to deal with all the data and were responsible for the decision to submit this manuscript for publication. YT, RL, YL, ML, YXS, YZ, AG and QW had the full data of the manuscript. YT, RL, YL, ML, and YXS were responsible for checking and evaluating the quality of the data and enrolled studies. YT was appointed for writing the draft of this manuscript. All authors contributed to the article and approved the submitted version.