Diagnostic accuracy of PET/CT scan in differentiating malignant and bland thrombi in cancer patients: A systematic review

Abstract Introduction: Differentiating between tumor thrombus and bland thrombus can be challenging, however it is of paramount importance as it can have implications on decisions pertaining to anticoagulation. PET/CT scans are routinely used in patients with cancer for staging and surveillance. Recently, they have also shown promise in distinguishing thrombi. In this review, we assess the diagnostic performance of PET/CT scans in differentiating between types of thrombi (bland versus tumor) in patients with cancer. Methods: We systematically searched PubMed and Embase for retrospective observational studies evaluating the diagnostic accuracy of PET/CT scans in differentiating tumor from bland thrombi in adult patients with cancer. We included studies involving patients who underwent subsequent confirmatory imaging, such as MRI, ultrasound or contrast- enhanced CT, histopathological evaluation, or clinical follow-up. Patients without a confirmed diagnosis of malignancy were excluded. Risk of bias assessment was done using QUADAS-2 tool. Diagnostic accuracy was assessed by pooling per-site (thrombus-level) data from individual studies, including true positives, true negatives, false positives, and false negatives. Sensitivity and specificity were estimated using a bivariate random-effects model, and results were visualized using a summary receiver operating characteristic (SROC) curve. Positive likelihood ratio (LR+), negative likelihood ratio (LR−), and diagnostic odds ratio (DOR) were calculated using random-effects models, with point estimates and 95% confidence intervals reported. Between-study heterogeneity was evaluated using the I² (I-squared) statistic. Results: Our search yielded 5,059 articles after the removal of duplicates. The extracted studies were reviewed independently by two authors and any discrepancies were addressed by a third author A.M.. Five studies met the prespecified inclusion criteria, encompassing a total of 326 patients (225 males and 101 females), with a mean age of 55.7 years. Most common malignancy described was renal cell carcinoma (67.2%) followed by hepatocellular carcinoma (18.1%) and cholangiocarcinoma (2.8%). Other malignancies included colorectal cancer, non-Hodgkin lymphoma, pancreatic cancer, lung cancer, and breast cancer. The most common sites involved were the renal, portal, and inferior vena cava veins. cases (273 thrombi; 79.1%), while the remaining cases were verified by imaging or clinical follow-up (72 thrombi; 20.9%). Four studies reported the mean SUVmax; the mean SUVmax for tumor thrombus was 5.8, compared to 2.2 for bland thrombus. The SUVmax cutoffs used by studies to differentiate tumor from bland thrombus ranged from 2.25 to 3.63. The pooled sensitivity across studies was 92.5% (95% CI: 78.9–97.6%), ranging from 71.4% to 100%. The pooled specificity was 97.6% (95% CI: 86.1–99.1%), ranging from 80% to 100%. The pooled negative likelihood ratio (LR−) was 0.096 (95% CI: 0.017–3.730), with moderate heterogeneity (I² = 27.2%). The pooled positive likelihood ratio (LR+) was 14.47 (95% CI: 4.92–42.58), also showing moderate heterogeneity (I² = 38.1%). The pooled diagnostic odds ratio (DOR) was 167.13 (95% CI: 23.82–1172.50), with substantial heterogeneity (I² = 67.1%). Conclusion: PET/CT scan may bebeneficial in distinguishing tumor and bland thrombus in patients with cancer and guide subsequent treatment, however, the wide confidence intervals and heterogeneity of available studies warrant caution when interpreting these results. Heterogeneity may be partially explained by the small sample sizes of some included cohorts, the range of distinct malignant conditions with varying FDG metabolic activity, and the lack of histopathological confirmation in approximately 20% of cases.