Diagnosis and management of p-CFL can sometimes be challenging, even for the most experienced neurosurgeons13. The presentation of clear rhinorrhea and/or headache is common in many conditions. Those that should be specifically considered as CFL are allergic rhinitis, common cold, vasomotor rhinitis, spontaneous intracranial hypotension, subarachnoid hemorrhage, and meningitis16, 17. CFL could also pose a serious hazard, and it is associated with delayed wound healing, meningitis, epidural infections, and pneumocephalus. These complications often lead to prolonged hospitalization, reoperation, and increased health care costs18–21. The most serious potential complication of CFL is meningitis22. Two patients of our study died from Klebsiella pneumoniae infection. Thus, it is vital to find an early evidence to diagnose CFL. In this study, we found that both the first-day postoperative NCCT pneumocephalus (NP) and the volume change of NP (the NP change) could predict p-CFL, and the latter has a higher predictive value. Compared with no NP patients, patients with no change and increased NP had a higher risk of having p-CFL, and 100% of increased NP patient were proved of p-CFL.
The possible pathophysiology for the correlation between the presence of NP and p-CFL can be explained by two theories: the inverted bottle mechanism and the ball-valve mechanism15. In the inverted bottle mechanism, it is postulated that as CSF flows out of the subarachnoid space through a dural-arachnoid tear, negative pressure is created within the subarachnoid space. The negative pressure prevents the leakage of more CSF, until air enters to take its place and equilibrates the pressure differential. In the second theory, the ball-valve mechanism hypothesizes that air enters through a fracture next to an air-containing space15. A vacuum drainage system predisposes a patient to a pneumocephalus in the presence of CFL15, 23,24.
We found that the NP change after the extraction of expansive sponge can better predict p-CFL than the first-day NP. Head NCCT examination was performed on the first day after operation routinely. At that time the patient's bilateral nasal cavity was filled with expansive sponge and was in a continuous lying state. Even if there was cerebrospinal fluid leakage, cerebrospinal fluid was not easy to flow out of the nasal cavity. Also it was difficult for air to enter the brain. Therefore, if CFL is slight during and after operation, pneumocephalus is not easy to appear in early postoperative CT. When the patients underwent off-bed training after extraction for 1 day, the second CT scan was performed. At this time, if CFL existed, CSF would be more likely to flow out than before. The reduced intracranial pressure makes air more likely to enter the brain and causes pneumocephalus. Therefore, detecting the changes of NP volume can more effectively predict p-CFL than observing the first NP. However, most of the patients discharged within 5 days after operation have made it difficult for us to do further CT dynamic observation.
In patients with the first-day NP after tumor resection, the occurrence of p-CFL was lower in patients with reduced NP volume comparing with those with increased or invariant NP volume. The probable cause is that pneumocephalus was gradually absorbed after the CFL was cured. CFL during operation and the loss of a large amount of CSF leading to pneumocephalus at first. However, after the leakage was properly repaired by skull base reconstruction, the postoperative pneumocephalus was gradually absorbed. In contrast, invariant and increase of CT pneumocephalus suggested a high incidence of CFL. If CFL is completely solved, pneumocephalus will be gradually absorbed, and CT re-examination at an interval of 3 days is enough to observe the reduction of pneumocephalus. If there is no change or increase of pneumocephalus volume after operation, it indicates that intracranial pressure continues to decrease and leakage still exists, so the incidence of p-CFL increases.
In this study, p-CFL happened in 2 patients who had no pneumocephalus on the first day after the operation. According to the dynamic evaluation, pneumocephalus was found in the head CT on the fourth day after the operation, suggesting that the dynamic review of head NCCT is of significance to the evaluation of CFL. In one case, although there was an increase in pneumocephalus volume after operation, there was no cerebrospinal fluid leakage, which may be related to the excessive drainage of cerebrospinal fluid caused by lumbar cistern drainage.
Our study is limited by the relatively small sample size, and further studies with larger sample size will be needed to confirm our results. An additional limitation is our retrospective and single-center methodology. Meanwhile, our observation time is short and limited to during hospitalization, requiring a longer follow-up.