The infiltration risk prediction models by logistic regression for ground-glass pulmonary nodules: a systematic review and meta-analysis.

METHODS

CNKI, Wanfang, VIP, Sinomed, Pubmed, Web of Science, Embase, and other databases were searched. The retrieval time was from the establishment of the database to January 31, 2024. We included all predictive models for the invasion of ground-glass pulmonary nodules established. The modeling group was patients with a pathological diagnosis of ground-glass pulmonary nodules. Two researchers screened the literature, established an Excel table for information extraction, used SPSS 25.0 to perform frequency statistics of each independent risk factor, and used Revman 5.4 software for meta-analysis.

RESULTS

A total of 29 articles were included, involving 30 independent risk factors, with a cumulative frequency of 99 times. There were 16 risk factors with a frequency of ≥2 times, a total of 85 times, accounting for 85.86%. The meta-analysis showed the following: average CT value (MD = 75.57 HU, 95%CI: 44.40-106.75), maximum diameter (MD = 4.99 mm, 95%CI: 4.22-5.77), vascular convergence sign (OR = 11.16, 95%CI: 6.71-18.56), lobulation sign (OR = 3.80, 95%CI: 1.59-9.09), average diameter (MD = 4.46 mm, 95%CI: 3.44-5.48), maximum CT value (MD = 112.52 HU, 95%CI: 8.08-216.96), spiculation sign (OR = 4.46, 95%CI: 2.03-9.81), volume (MD = 1,069.37 mm, 95%CI: 1,025.75-1,112.99), vacuole sign (OR = 6.15, 95%CI: 2.70-14.01), CTR ≥0.5 (OR = 7.24, 95%CI: 3.35-15.65), vascular type [types III and IV] (OR = 13.62, 95%CI: 8.85-20.94), pleural indentation (OR = 6.92, 95%CI: 2.69-17.82), age (MD = 4.18years, 95%CI: 1.70-6.65), and mGGN (OR = 3.62, 95%CI: 2.36-5.56) were risk factors for infiltration of ground-glass nodules. The overall risk of bias in the methodological quality evaluation of the included studies was small, and the AUC value of the model was 0.736-0.977.

CONCLUSION

The included model has a good predictive performance for the invasion of ground-glass nodules. The independent risk factors included in the model can help medical workers to identify the high-risk groups of invasive lung cancer in ground-glass nodules in time and improve the prognosis.