Performance of serum thymidine kinase 1 protein in lung cancer risk assessment: a systematic review and meta-analysis

This systematic review aims to update the differences in serum thymidine kinase 1 protein (STK1p) levels between tumor-free individuals, pneumonia patients, benign lung tumor patients, and lung cancer patients, as well as to provide a novel summary of the performance of STK1p-based models in lung cancer risk assessment. We systematically searched PubMed, Web of Science, and China National Knowledge Infrastructure (CNKI) for observational studies containing mean STK1p values with corresponding standard errors or area under the curves (AUCs) with 95% confidence intervals (CIs). We performed meta-analyses to estimate the pooled mean STK1p values, mean differences, and AUCs for predicting lung cancer risk. A total of 32 studies were eligible. The pooled STK1p levels in each group showed an increasing trend from tumor-free, pneumonia, benign and precancerous tumor, to lung cancer group. The pooled mean difference in STK1p level was significantly lower in tumor-free group compared to the pneumonia group (mean difference= -0.30[95% CI: -0.39,-0.21], p < 0.001, I 2 = 32.91%, n = 190). The tumor-free, pneumonia, and benign and precancerous tumor groups all had a lower STK1p compared to the lung cancer group. The pooled AUC for multivariate models including STK1p was 0.85 (95% CI: 0.79, 0.90, p < 0.001, I 2 = 74.57%, n = 7,020). Multivariate models incorperating STK1p and other predictors might be used for the risk assessment of lung cancer. An increase in model-predicted risk might advise doctors and patients to take further tests to diagnose the potential lesions in lung, but further prospective research is required.