Retrospective single-center study tests venous imaging strategy in mild-to-moderate brain injury

A retrospective study of 528 adults with mild-to-moderate traumatic brain injury suggests that a fracture crossing the anatomical course of a major dural venous sinus could serve as the first checkpoint for selecting patients who need dedicated venous imaging. Traumatic cerebral venous sinus thrombosis, or tCVST, was detected in 41 patients, or 7.8% of this preselected imaging cohort, and trans-sinus fracture was the strongest candidate risk marker examined. Patients with the fracture had 8.44 times the odds of tCVST, with a 95% confidence interval of 4.38 to 16.39.

The researchers then used D-dimer concentration and headache severity to stratify fracture-positive patients into groups with different imaging yields. Because the strategy was derived and evaluated in the same hospital cohort, the findings are exploratory and require prospective external validation before routine clinical use. The strategy addresses a complication that can be difficult to detect after head injury. tCVST occurs when trauma causes thrombotic occlusion of a dural venous sinus or large cerebral vein.

Mechanical injury to the sinus wall can disrupt the endothelium and initiate local coagulation, while obstruction of venous outflow may impair cerebrospinal fluid absorption, increase intracranial pressure and produce cerebral venous congestion. Thrombosis can extend to adjacent cortical veins, and elevated downstream venous pressure may contribute to venous infarction with hemorrhagic transformation. Standard non-contrast head CT and CT angiography designed to assess arterial injury may fail to visualize sinus thrombus, particularly with partial occlusion or subtle hyperdensity mistaken for imaging artifact.

Dedicated CT venography or MR venography provides a more direct assessment, but routine venous imaging is not standard in most traumatic brain injury protocols. Most research on tCVST has focused on severe brain injury, particularly cases with skull fractures near a venous sinus and patients requiring neurosurgical care. The study examined a less severely injured population: adults whose Glasgow Coma Scale score was above 8, corresponding to the mild-to-moderate range of 9 to 15.

Its first step was to inspect high-resolution CT bone-window images for a linear or comminuted fracture line traversing the anatomical course of the superior sagittal, transverse, sigmoid or straight sinus. A neuroradiologist blinded to venographic findings and the final tCVST diagnosis classified 100 patients as fracture-positive and 428 as fracture-negative. Among patients without the fracture, 16 of 428 had tCVST, while 25 of 100 fracture-positive patients had the complication.

That distribution made trans-sinus fracture a potentially useful gatekeeper but an incomplete risk marker on its own. It identified 25 of the 41 tCVST cases, giving 61.0% sensitivity and 89.1% specificity, while its negative predictive value was 96.3%. The positive likelihood ratio was 3.96, and the negative likelihood ratio was 0.46.

However, 16 cases occurred without a visible trans-sinus fracture, showing that the absence of this fracture would not exclude tCVST in every patient. Conversely, most fracture-positive patients did not have tCVST, meaning that imaging everyone with such a fracture would still produce many negative examinations. The authors therefore proposed using the fracture as an initial filter, followed by risk stratification rather than as a stand-alone diagnostic finding or automatic imaging command.

The second step assessed two readily available measurements only in fracture-positive patients. A patient was classified as high risk if D-dimer was at least 3.95 mg/L fibrinogen equivalent units or headache intensity was at least 3 on a 0-to-10 scale; both measurements had to fall below those cutoffs for a patient to enter the lower-risk group. D-dimer was measured within 24 hours of admission before antithrombotic administration, and the 3.95 mg/L threshold was the median among fracture-positive patients rather than an independently validated clinical threshold.

Headache was recorded at admission with a 100-mm visual analog scale, with a score of 3 corresponding to 30 mm. This “either-or” rule classified 77 patients as high risk, 24 of whom had tCVST, a detection rate of 31.2%. The 23 lower-risk patients included one case, or 4.3%.

When the two variables were examined in separate combinations, the highest observed yield occurred among fracture-positive patients with both elevated D-dimer and a headache score of at least 3: 16 of 29, or 55.2%, had tCVST. By comparison, the group with low D-dimer and a headache score below 3 had one case among 23, and those with high D-dimer but a lower headache score had one among 21, or 4.8%. These observations describe how two clinical measurements refined the observed probability of thrombosis after an anatomical risk signal had been identified.

The largest numerical attraction of the strategy was its change in imaging efficiency. If all 528 patients had undergone venous imaging, 41 cases would have been detected with a number needed to image of 13, corresponding to 13 examinations for each case found. Imaging every fracture-positive patient would require four examinations per detected case.

Restricting imaging to the 77 high-risk patients reduced the number needed to image to three and avoided 451 examinations classified by the researchers as unnecessary, or 85.4% of the total imaging volume. This efficiency came with a cost: the complete two-step pathway detected 24 of 41 cases, corresponding to 58.5% sensitivity, and would therefore have missed 17 cases. The reported 96.3% negative predictive value for Step 1 was also calculated in a cohort with a 7.8% detection yield, so predictive values could differ in unselected trauma populations.

The cohort itself limits how the results can be interpreted. These were not consecutive patients with mild-to-moderate brain injury. They were 528 of 566 adults who had already undergone CT venography, MR venography or selected digital subtraction angiography because their treating neurosurgeons suspected possible venous injury.

The study therefore measures diagnostic performance within an imaging-enriched population, not the true incidence of tCVST among everyone with mild-to-moderate traumatic brain injury. It was conducted retrospectively at one tertiary hospital in Shaanxi, China, from January 2024 through January 2026. Forty-one of the 528 included patients had tCVST, and 16 of the 41 cases lacked trans-sinus fracture.

Treatment decisions, including anticoagulation, were not standardized, and long-term functional recovery, thrombus recanalization and hemorrhagic complications were not systematically assessed. Several questions must be answered before the approach could become a clinical rule. Prospective multicenter studies would need to enroll consecutive patients, standardize fracture assessment, D-dimer collection and venographic timing, and determine whether deferring imaging in fracture-negative patients is safe.

The 16 fracture-negative tCVST cases also require detailed characterization to identify additional risk signals not captured by the strategy. A cost-effectiveness analysis would need to compare the two-step approach with universal screening and current discretionary practice, while the data-derived D-dimer cutoff would require confirmation in independent populations. The two-step pathway was developed and evaluated in the same sample, so its sensitivity, specificity, predictive values and efficiency estimates are likely optimistic.

Until external validation establishes its performance and safety, the results remain a hypothesis-generating decision aid: a CT-visible trans-sinus fracture opens the pathway, while D-dimer and headache thresholds refine the selection of patients for dedicated venous imaging without replacing clinical judgment.

Subject of Research: Exploratory two-step selection of venous imaging for traumatic cerebral venous sinus thrombosis after mild-to-moderate traumatic brain injury

Article Title: Trans-sinus fracture as the gatekeeper: a two-step clinical strategy to guide venous imaging in mild-to-moderate traumatic brain injury

Article References: Zhan, J., Chen, X., Liu, B., Zhou, X., Xu, D., & Wu, Z. (2026). Trans-sinus fracture as the gatekeeper: a two-step clinical strategy to guide venous imaging in mild-to-moderate traumatic brain injury. Frontiers in Neurology, 17, Article 1901656. https://doi.org/10.3389/fneur.2026.1901656

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1901656

Keywords: traumatic brain injury, cerebral venous sinus thrombosis, trans-sinus fracture, CT venography, MR venography, D-dimer, headache severity, diagnostic imaging, clinical decision rule, retrospective cohort, Trans-sinus, fracture