|
||||||||||||||||||||
|
|
Cerebrospinal Fluid Procalcitonin—A Potential Biomarker for Post-Craniotomy Bacterial Meningitis
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.344643
Keywords: Aseptic meningitis, bacterial meningitis, biomarkers, postoperative fever, postoperative meningitis, procalcitonin
Postoperative fever and suspected infection are the leading causes of prolonged hospitalization and readmission in neurosurgical patients undergoing a craniotomy/craniectomy.[1],[2],[3] Among all causes of fever in post neurosurgery setting, early diagnosis of bacterial meningitis is of paramount importance as early initiation of antibiotic therapy in bacterial meningitis plays an important role in reducing mortality.[4] In a review of 3,401 patients undergoing craniotomies for non-trauma indications, it was observed that the incidence of postoperative bacterial meningitis (PBM, defined as cerebrospinal fluid (CSF) culture demonstrating bacterial growth) was 0.8% and nearly four times as many patients were diagnosed to have postoperative aseptic meningitis (PAM) based on CSF cytological and biochemical parameters and negative CSF culture.[3] Following neurosurgical procedures, cell counts in CSF, as well as other biochemical parameters, have poor sensitivity in discriminating PBM from PAM and are not reliable.[4],[5],[6],[7] Over the past two decades, serum procalcitonin (PCT) has been reported to have a predictive value in diagnosing systemic sepsis.[8],[9] Elevated serum PCT level is a good predictor of a bacterial infection and has been used to initiate early antibiotic therapy in adults and children presenting with a febrile illness. In children and adults with febrile illness and suspected meningitis, elevated serum and CSF PCT has been used to distinguish viral and pyogenic meningitis.[7],[9],[10],[11],[12],[13],[14],[15],[16],[17],[18],[19] Following trauma and post-surgery, PCT levels may be elevated in serum.[8] Serum PCT levels have, however, not been found to be elevated following a craniotomy.[11] Serum PCT level has been shown to be elevated but has been demonstrated to have a limited role in diagnosing PBM.[20],[21] Recently, CSF PCT level was reported to be elevated in patients with PBM.[22] Postoperative meningitis following neurosurgery being a localized infection, it is likely that CSF PCT levels may be elevated with serum levels being not elevated. We hence evaluated the diagnostic utility of PCT values in CSF in distinguishing patients with PBM from PAM.
The study protocol was approved by the Institutional Review Board. All patients with suspected meningitis within 30 days following craniotomy for nontraumatic neurosurgical indications (between August 2010 and August 2018) were included in this study. Patients who underwent the surgery following trauma were excluded as trauma could be a potentially confounding factor causing elevation in PCT. Patients who underwent prior intracranial surgery and shunt placement were also excluded. Any patient who presented with fever in the postoperative period with the clinical features of meningitis underwent CSF analysis. Normal CSF values were determined from CSF specimens collected during insertion of the lumbar subarachnoid drain or from ventricle tapped for CSF diversion procedure in patients with noninfectious indications with hydrocephalus. Definition of meningitis As per the Centers for Disease Control and Prevention (CDC) criteria, meningitis was diagnosed on a positive CSF stain or culture or alteration of cytochemical parameters.[23] PBM was defined as the presence of bacteria demonstrated in CSF on culture. PAM was diagnosed in patients with sterile CSF who had any one or more of the following on CSF analysis—elevated CSF, total WBC count, elevated protein, or low levels of sugar.[3] None of the patients with PAM had systemic infection. Categorization of patients Based on the above definition, patients were grouped into three: normal (n = 14), PBM (n = 28), and PAM (n = 31). The median age was 26 years (range: 2–62 years). CSF analysis The mean time since surgery for CSF analysis for suspected meningitis was 9.9 ± 6.7 days (range: 1–28 days). All samples of CSF were analyzed for total and differential WBC count, protein, sugar, and bacterial cultures. CSF was deemed to be sterile if no growth was noted for 7 days after the culture plates were incubated. CSF PCT assay PCT was measured in CSF using a fluorescence homogenous immunoassay named TRACE (time-resolved amplified cryptate emission) on Kryptor compact instrument (Thermo Fisher Scientific). PCT in the patient's sample was sandwiched between two antibodies, one labeled to a donor Europium tripyridyl cryptate and another to the light-absorbing algal protein—the acceptor. The formation of the immune complex results in light from a nitrogen laser (620 nm) being absorbed by the acceptor, which then emits the light at 665 nm in milliseconds. In the unbound form, the same emission occurs in nanoseconds, thus resolving the free and bound form. The measuring range for this assay was 0.02–40 ng/mL. Concomitant serum PCT assay using the same technique was performed in 17 of the 28 patients with PBM and 12 of the 31 patients with PAM. Statistical analysis Descriptive statistics were obtained for all the CSF parameters that were analyzed. As the distribution of CSF PCT in all the groups did not follow a normal distribution, nonparametric Kruskal Wallis test was used to study difference between the various parameters between three groups. The comparison of CSF PCT levels between groups was performed using Mann-Whitney U test. The P value of <0.05 was considered significant. The receiver operating characteristic (ROC) curve was constructed for CSF PCT values. The sensitivity, specificity, positive and negative predictive values, and likelihood ratios were calculated.
CSF cell counts, sugar, and protein As summarized in [Table 1], there was no significant difference between CSF total WBC count, sugar, and protein values between the PAM and PBM group. These parameters were significantly different in PAM and PBM group when compared to normal values.
CSF PCT The median PCT level in CSF in the normal group was 0.03 ng/mL (IQR 0.02 to0.06ng/ml). In the PAM group, the median PCT level in CSF was 0.16 (IQR 0.09-0.3ng/ml) while it was 0.37 ng/mL (IQR 0.2 to 1.4 ng/mL) in the PBM group. There was a statistically significant elevation of CSF PCT in PBM group when compared to normal (P < 0.001) and PAM group (P = 0.0004) as shown in [Table 1]. The area under the ROC curve for CSF PCT was 0.767 (P = 0.01, 95% CI: 0.641–0.892) [Figure 1]. A cutoff value of ≥0.12 ng/mL yielded a sensitivity of 85.7% and specificity of 51.6% for the diagnosis of PBM. The positive predictive value of the test was 61.5% and the negative predictive value was 80%. The positive likelihood ratio was 1.8 while the negative likelihood ratio was 0.3. The test metrics are summarized in [Table 2].
Serum PCT levels In 17 of the 28 patients in the PBM group, CSF and serum were tested simultaneously for PCT. Of the 10 patients with serum PCT levels of 1 ng/mL or more, there was evidence of a systemic infection such as pneumonia or urinary tract infection or positive blood culture. None of the seven patients whose serum PCT was <1 ng/mL had evidence of a systemic bacterial infection. In the PAM group, 12 of the 31 patients had simultaneous serum PCT estimation and it was elevated in two of these patients. Overall, no correlation could be established between the serum and CSF PCT values in either PAM or PBM group.
Among the etiologies for postoperative fever after neurosurgery, PBM is the most life-threatening of all infections and needs to be recognized promptly as early initiation of antibiotic therapy is important in reducing mortality from the same. It has been recognized that post-surgery, a significantly higher number of patients develop PAM than PBM.[3] The role of serum PCT, CSF WBC counts, and biochemical parameters in diagnosing PBM or distinguishing it from PAM following surgery is poor.[4],[5],[24],[25] In our patient cohort, all patients with PBM had positive CSF cultures and those diagnosed as PAM had no identifiable systemic infections and their CSF cultures were sterile. In this cohort too, none of the CSF cytological or biochemical parameters could separate PBM from PAM. Serum PCT in postsurgical bacterial meningitis As the level of serum PCT has been reported to remain <0.2 ng/mL during the first-week post-neurosurgery in patients without any infection, it has a potential in predicting PBM and ventriculitis post external ventricular drain insertion.[24],[26],[27] In a recent study, the sensitivity of serum PCT value of ≥0.15 ng/mL in independently detecting patients with PBM (all 14 patients had positive CSF culture) was only 50%, although the specificity was 80%.[21] The sensitivity could not be improved by combining the criteria of elevated C-reactive protein and elevated peripheral white blood cell count with serum PCT ≥0.15 ng/mL although the specificity improved to 92.6% with this combination.[20],[21] Contrary to this, Yu et al.[28] reported high sensitivity and specificity for serum PCT (actual values not mentioned) for diagnosing post-neurosurgical intracranial infection (only 54.3% of whom had positive CSF culture). Serum PCT level of 1 ng/mL has been reported to be diagnostic of ventriculitis, with positive CSF culture, with sensitivity of 68% and specificity of 77%.[27] Elevated serum PCT level has been documented in ventriculitis (post external ventricular drain insertion) even with bacteria of low virulence.[24] In our patient cohort, synchronous bacteremia, respiratory, or urinary tract infections were detected in all 10 of 17 patients with PBM who had serum PCT >1 ng/mL. In the seven patients with serum PCT <1 ng/mL, there was no systemic infection. From these observations, serum PCT level alone may not be discriminatory in the case of isolated PBM in the absence of a systemic infection. No correlation between serum and CSF PCT could be established in the subgroup of patients who underwent synchronous serum and CSF PCT assays. CSF PCT in postsurgical bacterial meningitis Li et al.[22] reported that the CSF PCT level of 0.075 ng/mL or more could aid in diagnosing PBM (sensitivity, 68%; specificity, 72.7%) among 178 patients, 50 of whom were diagnosed to have PBM (only 6–8% had positive CSF culture). Alons et al.[29] reported elevated CSF PCT levels in patients with PBM. However, only 3 of their 10 patients had positive CSF culture. Yu et al.[28] observed that the diagnostic sensitivity and specificity of CSF PCT (though actual values not reported) was higher than other indicators in patients with suspected post craniotomy meningitis. However, only 19 of the 35 patients in their “intracranial infection” group had positive microbial cultures and thus 16 of their patients could have had aseptic meningitis rather than PBM. Liu et al.[25] reported serial measurements of PCT and other inflammatory cytokines in CSF from patients with postoperative external ventricular drains. They detected significant elevation in CSF PCT only on the 4th day after the onset of fever in eight patients with positive CSF cultures while the levels were not elevated in 11 patients with aseptic meningitis. The sensitivity and specificity of CSF PCT was 62.5% and 80%, respectively. In our patient cohort, CSF PCT levels in 28 patients with PBM (in the setting of neurosurgery for non-trauma indications and all proven by positive CSF culture) were significantly elevated compared to those in normal individuals as well as patients with PAM. The performance of this test with a cutoff value of ≥0.12 ng/mL yielded a sensitivity of 85.7% and specificity of 51.6%. Our cutoff value has a higher sensitivity though lower specificity than reported by Li et al.[22] We demonstrated that elevated CSF PCT was useful in discriminating between PAM and PBM, with a negative predictive value of 80% and a positive likelihood ratio of 1.8. The results of studies reporting CSF PCT in PBM have been summarized in [Table 3]. In 4 among the 28 patients with PBM, postoperative CSF PCT levels during an episode of fever remained normal when the concomitant CSF culture was sterile and was elevated subsequently when CSF culture was positive. Elevated PCT in our patients with PBM is thus truly predictive of bacterial meningitis and the elevation is not a result of changes in the CSF due to the surgical procedure.
Use of other CSF markers Several authors have reported elevated CSF lactate >4 mmol/L concentrations to be useful in discriminating bacterial from aseptic meningitis in patients with suspected post neurosurgical meningitis.[22],[30],[31],[32],[33] Li et al.[22] reported that the sensitivity of CSF PCT was lower than that of CSF lactate in their patient cohort with PBM. A parallel test using a paradigm of CSF lactate and CSF PCT could increase the sensitivity to 96%, compared with either marker alone.[22] Contrary to these reports, CSF lactate was found to have poor discriminatory value in predicting bacterial infection (definition of bacterial meningitis was not by CSF culture and was presumptive) though serial fall in its levels correlated with improvement in the clinical status as well as other CSF biochemical parameters.[6] Similarly, Berger et al.[24] detected no difference in CSF lactate levels in patients with and without a positive CSF culture when ventriculitis was suspected. Liu et al.[25] reported that values of inflammatory cytokines such as tumor necrosis factor-α, interleukin-1β, and interleukin-8 were elevated in CSF 4 days prior to the onset of fever in patients with PBM when compared to the patients with PAM. In comparison, CSF PCT levels were elevated significantly on day 4 after the onset of fever. However, the results of this study may not be translated into a clinical setting where serial CSF analysis is not feasible or cost-effective in every patient with external ventricular drain. Limitations As the study was designed to collect data on CSF PCT prospectively and then analysis was done, serum PCT levels were not available for all patients in the PAM/PBM groups. With the available data, no correlation could be detected between serum and CSF PCT levels in either group. The cost of performing CSF PCT level would be 1,100 INR. Though the PPV of this test was 61.5%, it had a higher negative predictive value of 80% and could rule out PBM. As all patients were treated with antibiotics until CSF cultures were obtained, the cost-effectiveness of this test could not be assessed. Future studies designed to prospectively assess the cost-effectiveness of this test need to be done. CSF was plated in an aerobic media only in our patient cohort and reported to be sterile if no growth was observed for 7 days. As CSF was not plated into anaerobic media, low virulent organisms such as Propionibacterium acnes were not reported in any sample. In none of the culture plates with aerobic media, there was suspicion of growth of P. acnes. However, post neurosurgical P. acnes infection has been reported to be very indolent and often associated with extradural abscess, subgaleal abscess, and osteomyelitis rather than isolated meningitis.[34],[35],[36] The median duration of infections secondary to P. acnes was more than 30 days in most of the cases reported in literature. Considering the time interval between cranial surgery and P. acnes isolation in the literature, it is unlikely that any of our patients would have harbored this organism. Although we did not withhold antibiotics based on the value of CSF PCT in any patient with suspected PBM, antibiotics were discontinued when CSF culture was negative after 72 h of incubation. To establish guidelines, a prospective study on the utility of this test on decision-making regarding diagnosis and initiation of treatment needs to be designed.
CSF PCT assay in patients with suspected meningitis within 30 days of a neurosurgical intracranial procedure for non-trauma indications has a role in discriminating PBM from PAM while awaiting CSF culture reports. Acknowledgements This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. No material support was received for this research. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.
[Figure 1]
[Table 1], [Table 2], [Table 3]
|
|
|||||