on 09-Mar-2021 (Tue)

Patients with tuberculous meningitis commonly present with stiff neck, headache, fever, and vomiting; these symptoms are also frequently observed with bacterial meningitis

Features that may help distinguish tuberculous meningitis from bacterial meningitis include:

● Subacute presentation; in one study including 160 patients with tuberculous meningitis, the time between the onset of symptoms and clinical presentation was less than one week in 7 percent of cases, 1 to 3 weeks in 57 percent, and more than 3 weeks in 36 percent. In contrast, bacterial meningitis typically presents within one week of symptoms; potential exceptions include meningitis due to listeriosis or brucellosis. (See "Clinical features and diagnosis of acute bacterial meningitis in adults".)

● Presence of neurologic symptoms; in the above study including 160 patients with tuberculous meningitis, altered consciousness, personality changes, and coma were observed in 59, 28, and 21 percent of cases, respectively [2]. Such symptoms are not typical of bacterial meningitis. (See "Clinical features and diagnosis of acute bacterial meningitis in adults".)

● Presence of cranial nerve palsies (most frequently involving cranial nerve II and VI) are common; other cranial nerves are infrequently involved. In one study including 158 patients with tuberculous meningitis, cranial nerve palsies were observed in almost one-third of cases. Cranial nerve palsies are uncommon with bacterial meningitis, exceptions include meningitis due to listeriosis or meningitis associated with intracranial abscess [3].

Patients with tuberculous meningitis typically progress through three discernible phases [4-7]:

● The early prodromal phase, lasting one to three weeks, is characterized by the insidious onset of malaise, lassitude, headache, low-grade fever, and personality change.

● The meningitic phase follows with more pronounced neurologic features, such as meningismus, protracted headache, vomiting, lethargy, confusion, and varying degrees of cranial nerve and long-tract signs.

● The paralytic phase supervenes as the pace of illness accelerates rapidly; confusion gives way to stupor and coma, seizures, and often hemiparesis. For the majority of untreated patients, death ensues within five to eight weeks of the onset of illness.

Clinical presentation with central nervous system (CNS) manifestations in the absence of prior history of pulmonary symptoms is common; in one series including 61 patients with tuberculous meningitis, history of tuberculosis (TB) disease was elicited in only 10 percent of cases [8].

Patients may present with atypical features that mimic other neurologic conditions. These include an acute, rapidly progressive meningitic syndrome suggesting pyogenic meningitis, or with slowly progressive dementia over months characterized by personality change, social withdrawal, loss of libido, and memory deficits. Less commonly, patients may present with an encephalitic course manifested by stupor, coma, and convulsions without overt signs of meningitis [11].

Hydrocephalus occurs in up to 80 percent of patients with tuberculous meningitis, and is often accompanied by signs of raised intracranial pressure. Hydrocephalus should be suspected when patient has features of elevated intracranial pressure, deteriorating vision, and/or deteriorating consciousness. In such cases, neuroimaging should be performed urgently [12]

Hyponatremia is a potentially serious complication of tuberculous meningitis. Hyponatremia may develop at any point of time during course of tuberculous meningitis. In one study including 76 cases, hyponatremia was observed in approximately 45 percent of cases [14]. The onset and progression can be insidious with nonspecific symptoms (nausea, confusion, delirium, seizure) that may be mistakenly attributable to other pathologic features of the infection or its treatment.

Assessment of volume status is important to differentiate patients with the syndrome of inappropriate anti-diuretic hormone secretion (SIADH) from patients with central salt wasting; in general, patients with SIADH are euvolemic, while patients with central salt wasting are hypovolemic.

Vision loss – Vision loss, a severely disabling complication of tuberculous meningitis, occurs in approximately one-quarter of patients [15,16]. Mechanisms of vision loss include optochiasmatic arachnoiditis (encasement of optic nerves and optic chiasma by thick tuberculous exudates), compression of the optic chiasma by the dilated third ventricle from above, elevated intracranial pressure, endarteritis of vessels supplying the optic nerves and optic chiasma, and direct M. tuberculosis invasion of the optic nerves. Many survivors of tuberculous meningitis have permanent blindness.

Among patients with HIV infection, the presentation of tuberculous meningitis is the same as in patients without HIV infection; however, concurrent TB outside the CNS and lung occurs more commonly. In addition, initiation of antiretroviral therapy in patients with HIV infection and no prior CNS symptoms may precipitate onset of new symptoms, reflecting progression of underlying TB infection as a manifestation of immune reconstitution inflammatory syndrome.

Routine blood counts and chemistries are relatively nonspecific; mild anemia is common. Hyponatremia may be observed in the context of inappropriate antidiuretic hormone or cerebral salt wasting [14,17]. Central diabetes insipidus resulting in hypernatremia has also been described [18]

CT of the chest is an important diagnostic modality to evaluate for concomitant pulmonary tuberculosis (image 1). In one study including 81 patients with tuberculous meningitis, chest CT thorax abnormalities were observed in two-thirds of patients; centrilobular nodules were the most common lung parenchyma abnormality, and asymptomatic miliary lung lesions were observed in 10 percent of tuberculous meningitis cases [19]

The diagnosis may be definitively established in the setting of cerebrospinal fluid (CSF) with positive smear for acid-fast bacilli (AFB), CSF culture positive for M. tuberculosis, or CSF with positive nucleic acid amplification test (NAAT) [21]. However, definitive diagnosis can be challenging, given suboptimal sensitivity and specificity of diagnostic tests [20,22]

A presumptive diagnosis of tuberculous meningitis may be made in the setting of relevant clinical and epidemiologic factors and typical CSF findings (lymphocytic pleocytosis, elevated protein concentration, and low glucose concentration).

CSF examination – At the time of lumbar puncture (LP), the opening pressure should be measured. CSF should be sent for total cell count with white blood cell differential, protein concentration, and glucose concentration (with serum level for comparison), NAAT, and AFB smear and culture; in patients with HIV infection, CSF testing should also include cryptococcal antigen. Additional fluid should be saved in case additional studies are warranted

The diagnostic yield may be increased by submitting multiple CSF specimens (at least three) from repeated LPs and greater volumes (at least 5 mL); however, empiric antituberculous therapy should not be delayed while obtaining serial CSF specimens [24]. The sensitivity of AFB smear and culture is maintained for a few days after starting treatment, and the sensitivity of CSF NAAT is maintained for up to one month after starting treatment [17,25]

For patients with mild-to-moderate hydrocephalus in the absence of signs of elevated ICP, LP may be pursued. For patients with contraindications to immediate LP in the setting of characteristic clinical, epidemiologic, and radiographic features of CNS TB, empiric therapy should be initiated promptly (and not be deferred pending LP).

Patients with suspected or proven tuberculous meningitis should always be tested for HIV infection.

Among patients with suspected CNS TB, a careful neurologic examination should be undertaken, including assessment of mental status, cranial nerves, sensory and motor examination, cerebellar function, and reflexes.

In addition, patients should be evaluated for other systemic manifestations of TB. Signs of active TB outside the CNS are of diagnostic importance if present. Patients should be assessed for fever, weight loss, jaundice, lymphadenopathy, hepatomegaly, splenomegaly, bone and joint lesions, and dermatologic findings. In patients with miliary TB, careful funduscopic examination often demonstrates choroidal tubercles; these are multiple ill-defined raised yellow-white nodules of varying size near the optic disk (picture 1).

Routine studies — The opening pressure is usually moderately elevated (180 to 300 mm H₂O) [18]. Typical CSF findings in tuberculous meningitis include lymphocytic pleocytosis, elevated protein concentration, and low glucose concentration (table 2).

● The CSF white blood cell count is typically between 100 and 500 cells/microL, with lymphocyte predominance. Early in the course of illness, a lower cell count and/or neutrophilic predominance may be observed. In addition, neutrophilic predominance may be observed in the setting of spinal block and/or paradoxical worsening.

● The CSF protein concentration is typically 100 to 500 mg/dL. However, in patients with subarachnoid block, the CSF protein concentration may be as high as 2 g/dL [17,18]

● The CSF glucose concentration is <45 mg/dL in most cases.

The sensitivity of acid-fast staining for diagnosis of tuberculous meningitis is 30 to 60 percent [24,26]. The diagnostic yield is increased with volume of CSF (up to 10 to 15 mL) and number of CSF specimens (up to four) [5,24]. AFB smears and cultures may be positive even days after treatment has been initiated

Organisms can be demonstrated most readily in a smear of the clot or sediment. If no clot forms, addition of 2 mL of 95% alcohol induces a heavy protein precipitate that carries bacilli to the bottom of the tube upon centrifugation

The sensitivity of CSF AFB culture for diagnosis of tuberculous meningitis is often <50 percent, although some studies report positive culture rates of up to 85 percent [5,24].

In one meta-analysis including 63 studies comprising more than 1300 patients with tuberculous meningitis (confirmed by culture), the specificity and sensitivity of NAATs were 99 percent (95% CI 98-99) and 82 percent (95% CI 75-87 percent), respectively [27].

A negative Xpert MTB/RIF test result should not be used to exclude tuberculous meningitis, given variable sensitivity [22]. Mycobacterial DNA may remain detectable in CSF for up to a month after initiation of treatment [25].

In low-prevalence settings, the Xpert MTB/RIF assay may be used as diagnostic tool if the assay has been validated by the laboratory for this indication [40]. In one systematic review and meta-analysis including 18 studies, the sensitivity and specificity for this assay in CSF (compared with culture) were 81 and 98 percent, respectively [34]. In addition, the positive predictive value of NAATs is limited in low-prevalence settings [35]

Quantitative NAAT results may be a useful prognostic tool; a high bacterial load correlates with cycle threshold (Ct) <16, medium load with Ct 16 to 22, low load with Ct 22 to 28, and very low load with Ct >28. A high CSF M. tuberculosis bacterial load prior to treatment (measured using the Xpert MTB/RIF assay) has been associated with higher disease severity and higher likelihood of new neurologic events after initiation of therapy [41].

Measurement of the CSF adenosine deaminase (ADA) level may be a useful adjunctive test for diagnosis of tuberculous meningitis [23,42]. However, elevated CSF ADA level may also be observed in the setting of bacterial infections and neurobrucellosis [42,43], and there is no clear threshold to distinguish tuberculous meningitis from meningitis caused by other infectious agents.

One meta-analysis included 10 studies and more than 350 with tuberculous meningitis (most of which defined an elevated ADA as 9 or 10 U/L) estimated the sensitivity and specificity of ADA for diagnosis of tuberculous meningitis to be 79 and 91 percent, respectively [44]. Another meta-analysis including 13 studies and more than 1000 patients with tuberculous meningitis noted the sensitivity and specificity of ADA for diagnosis of tuberculous meningitis depended on the definition of an elevated ADA level [45]. For ADA threshold of 4 U/L, the sensitivity and specificity were >93 and <80 percent, respectively; for ADA threshold of 8 U/L, the sensitivity and specificity were <59 and >96 percent, respectively.

In patients with tuberculous meningitis, classical neuroimaging findings include hydrocephalus, basilar exudates, periventricular infarcts, and cerebral parenchymal tuberculomas (image 2 and image 3 and image 4 and image 5 and image 6) [18,22]. Presence of basilar meningeal enhancement along with presence of hydrocephalus is strongly suggestive of tuberculous meningitis (image 6) [50-53].

In the setting of stage I disease, the CT scan is normal in approximately 30 percent of cases. In the setting of stage III disease, CT findings may include hydrocephalus as well as basilar enhancement. Marked basilar enhancement has been correlated with presence of vasculitis and risk for basal ganglia infarction

In two large community-based series, hydrocephalus was observed in approximately 75 percent of cases, basilar meningeal enhancement in 38 percent, cerebral infarcts in 15 to 30 percent, and tuberculomas in 5 to 10 percent [50,51]

Optochiasmatic arachnoiditis, an important cause of vision loss, is characterized radiographically by presence of tuberculous exudates in the region of interpeduncular, suprasellar, perimesencephalic, and Sylvian cisterns [54].

Periventricular brain infarcts are seen in up to half of patients and are common in advanced disease. Most infarcts are observed in the basal ganglia. In many patients, infarcts are asymptomatic and not associated with focal neurologic deficits [55,56].

Fungal meningitis (cryptococcosis, histoplasmosis, blastomycosis, coccidioidomycosis) – As with tuberculous meningitis, fungal meningitis often presents subacutely and may be associated with cranial neuropathies. Immunosuppression (due to HIV infection or other causes) is an important risk factor. In patients with cryptococcal meningitis, the CSF profile classically demonstrates elevated opening pressure, low white blood cell count with a mononuclear predominance, slightly elevated protein concentration, and low glucose concentration. Definitive diagnosis is made by CSF culture; a positive cryptococcal antigen in the CSF or serum strongly suggests the presence of infection.

Neurobrucellosis – Brucellosis typically presents with insidious onset of fever, malaise, night sweats, and arthralgias. It may be complicated by neurobrucellosis; manifestations include meningitis, encephalitis, brain abscess, myelitis, radiculitis, and/or neuritis (with involvement of cranial or peripheral nerves). CSF findings include a pleocytosis (10 to 200 white blood cells, predominantly mononuclear cells), mild to moderately elevated protein levels, hypoglycorrhachia, and elevated adenosine deaminase. The diagnosis may be established via antibody or agglutination testing of spinal fluid; uncommonly, the organism may be recovered in CSF culture.

Neurosyphilis – Meningitis and meningovascular disease are manifestations of early neurosyphilis. Clinical manifestations of syphilitic meningitis include headache, confusion, nausea, vomiting, and stiff neck; in addition, cranial neuropathies may occur. Visual acuity may be impaired if there is concomitant uveitis, vitreitis, retinitis, or optic neuropathy. In addition, an infectious arteritis affecting the vasculature of the subarachnoid space may occur, result in thrombosis, ischemia, and stroke involving the brain or spinal cord. Diagnostic testing includes serum treponemal and nontreponemal tests and spinal fluid examination to assess for presence of pleocytosis, elevated protein concentration, and CSF Venereal Disease Research Laboratory tests.

Bacterial meningitis – Bacterial meningitis presents acutely with fever, nuchal rigidity, and altered mental status. Typical CSF findings include white blood cell count of 1000 to 5000/microL (neutrophil predominance), protein concentration 100 to 500 mg/dL, and glucose concentration <40 mg/dL. The diagnosis may be established via isolation of a bacterial pathogen from CSF or blood cultures. In patients with partially treated meningitis, there is usually minimal effect on the CSF chemistry and cytology findings, but the yield of Gram stain and culture may be diminished.

Focal parameningeal infection (brain abscess, spinal epidural abscess, sphenoid sinusitis) – Clinical manifestations of parameningeal infection may be subacute and nonspecific. Initial evaluation should consist of radiographic imaging (of the brain and/or spinal cord). LP is contraindicated in patients with papilledema or focal symptoms or signs; if it is feasible to obtain CSF, findings may demonstrate elevated protein concentration, low glucose concentration, and pleocytosis. A microbiologic diagnosis may be established culture of material obtained via stereotactic-guided aspiration or surgery.

Viral meningitis (herpes simplex, mumps) – Viral CNS infection usually presents acutely, in contrast to tuberculous meningitis, which is typically subacute. In the setting of viral meningitis, the CSF also typically demonstrates a lymphocytic pleocytosis, elevated protein but normal glucose, and negative Gram stain and culture. Definitive diagnosis of viral meningitis is established via CSF polymerase chain reaction.

Neoplastic meningitis – Leptomeningeal metastases (LM) are diagnosed in approximately 5 percent of patients with advanced cancer; the most common primary tumors associated with development of LM are breast, lung, and melanoma. Clinical manifestations include headache, nausea and vomiting, leg weakness, ataxia, altered mental status, diplopia, and facial weakness. Typical MRI findings include linear and nodular leptomeningeal enhancement; thickening and enhancement of cranial nerves or nerve roots, including the cauda equina, and hydrocephalus. CSF findings include elevated opening pressure, mild pleocytosis, elevated protein and decreased glucose concentrations. The diagnosis of LM is definitively established via identification of malignant cells in the CSF.

The work of Bisagni and Lanzi [23] recognises the efforts made by other authors to perform a minimum weight optimisation of a stiffened composite panel. Boundary conditions and outputs often included minimum panel strength and the critical buckling load. Inputs include the layup definition as well as stiffener count and dimensions, which means both discrete and continuous variables are involved. To cope with such variables as well as a potential non-convex cost function, a genetic algorithm (GA) is chosen as optimisation algorithm. As this approach is computationally expensive, Bisagni and Lanzi propose an optimisation procedure in which the desired outputs of the structural response are predicted by means of a system of ANNs as shown in Figure 2.12. These networks are first trained on a series of FEM simulations, after which the calculation can be omitted in the optimisation process itself. A separate ANN is trained for each output variable: Pre-buckling stiffness, buckling load, collapse load and even the load-displacement curve. Each ANN featured a relatively simple architecture with only one HL and a limited number of nodes. The only ANN with more than 10 nodes is the load-displacement curve predictor which is intuitive due to the various inputs such a curve depends on. The networks consistently predicted the desired results with more than 75% of training and test data within a 4% error margin. The improvement in computational time of this approach in comparison with an FEA-only optimisation is around an order of magnitude.

Multiple demonstrations of ANN constitutive models are presented in Javadi et al. [33]. Similar to the work of Ghaboussi et al., the ANN is trained to predict the stress-strain behaviour based on the current state and a given increment size. Javadi et al. implemented a trained ANN in each element in an FE model as a substitution for the conventional constitutive material model. Apart from the computational procedure in each element, the analyses are performed by a conventional FE solver. The approach is tested on a 2-D beam with 10 elements and a transverse end-load, where the ANN is shown to be as accurate as the regular FE analysis. Furthermore, a quarter segment of a 2-D thick cylinder shell with nine 8-node elements subject to a distributed load is investigated. Again, an excellent agreement with both theoretical and FEM results is achieved with the ANN constitutive model. Finally, an analysis is performed involving the more complex

constitutive behaviour of an embankment consisting of a non-linear Mohr-Coulomb soil material. The ANN, trained on triaxial experimental tests results, is shown to be potentially more robust than an FEA as it is based directly on the experiments instead of a mathematical model that is subject to assumptions.