Systematic Review & Meta-Analysis of Positron Emission Tomography/Computed Tomography and Bone Scan in the Diagnosis of Prostate Lesions

Jiafu Wang; Yue Han; Lin Lin; Linhan Zhang; Jin Li; Huiqi Gao; Peng Fu


Transl Androl Urol. 2021;10(11):4231-4240. 

In This Article


Literature Search

The retrieval strategy developed by The Cochrane Collaboration was executed by combining subject headings and free words in the following databases: China National Knowledge Infrastructure (CNKI) database, Wan Fang Medical Network, PubMed, Excerpta Medica dataBASE (EMBASE), Medline, and the Cochrane database. We searched for articles on the diagnostic performance of 68Ga-PSMA PET/CT and bone scanning in the detection of prostate lesions, published between the establishment of each database and 20 April 2021. The search keywords in Chinese and English included 68Ga-PSMA, PET/CT, prostate lesions, prostate adenocarcinoma, bone metastasis, and BS. The quality of the literature was evaluated according to the RevMan 5.3 software provided by The Cochrane Collaboration. Will be free combination, retrieval words mentioned above, after many times to retrieve access can be included in the references, and the retrieved references to potential eligibility screening, and then use a search engine to trace of incompletion of literature, eventually contact with relevant experts, researchers, published literature the latest research progress.

Literature Inclusion and Exclusion Criteria

Inclusion criteria: (I) domestic or international publication involving 68Ga-PSMA PET/CT or BS imaging of prostate lesions; (II) patients with prostate lesions confirmed by histopathology and 68Ga-PSMA PET/CT imaging data; (III) comparative analysis of diagnostic results and pathology showed that the index was reliable at 95% confidence interval (CI); (IV) the research assumptions and methods of the studies are similar, and there is a clear publication period.

Exclusion criteria: (I) subjects with non-thyroid prostate disease; (II) repeated published studies, case reports, lectures, and reviews; (III) unable to contact the original author of the literature where complete data is not available.

Outcome Indicator

Studies must include: true positive (TP), true negative (TN), false positive (FP), false negative (FN), sensitivity, and specificity.

Data Extraction

Two evaluators used Microsoft Excel (Microsoft, the United States) to independently conduct literature screening and data extraction, and then cross-check. Any disagreement was resolved through discussion. The key extracted data included: (I) the basic information included in the study: the title of the study, the name of the first author, the publication year, the publication journal; (II) study sample description: sex, age, number of cases, pathological nature of cancer, size of lesion, etc.; (III) study design, scanning mode, and reference standards; (IV) number of TP, FP, FN and TN results.

Quality Evaluation and Biased Risk Assessment

The risk of bias in the included randomized controlled trials was assessed by two researchers at the same time, and the results were determined by discussion if the two disagreed. For the included references, the quality of the method was evaluated using the QUADAS2 quality assessment tool from the accuracy study. The tool consists of four neighborhoods which discuss case selection, trial to be evaluated, gold standard, and case process/progress, respectively. All areas were assessed based on the risk of bias and rated as "high risk", "low risk" or "unclear". 68Ga-PSMA PET/CT or BS was considered the "test to be evaluated", and histopathological analysis results are designated as the "gold standard". Table 1 shows the basic contents of the QUADAS2 quality assessment tool.

Statistical Analysis

Statistical analysis was performed using Stata SE12.0 software (Stata Corporation, USA). The odds ratio (OR) was used as the dichotomous variable. The risk bias of the included references was assessed using the bias risk assessment graph facility in the Rev Man 5.3 software. The data were sorted, screened, and input into the RevMan 5.3 software. Results and charts were obtained after analysis. All effects were expressed with 95% CI. When P>0.01 and I2<50%, the fixed effects model was adopted. When P<0.01 and I2>50%, the random effects model was adopted. The sensitivity and specificity of the two diagnostic modalities were calculated, and the summary receiver operating characteristics (SROC) was plotted for the two diagnostic modalities.