Performance of Lipoarabinomannan Assay using Cerebrospinal fluid for the diagnosis of Tuberculous meningitis among HIV patients

Background: The diagnostic utility of the Mycobacteria tuberculosis lipoarabinomannan (TB-LAM) antigen lateral flow assay on cerebrospinal fluid (CSF) for the diagnosis of tuberculous meningitis (TBM) has not been extensively studied and the few published studies have conflicting results. Methods: Lumbar CSF from 59 HIV-positive patients with suspected TBM was tested with TB-LAM and Xpert MTB/Rif Ultra. The diagnostic performance of CSF TB-LAM was compared to positive CSF Xpert MTB/Rif Ultra (definite TBM) and a composite reference of probable or definite TBM according to the uniform case definition. Results: Of 59 subjects, 12 (20%) had definite TBM and five (9%) had probable TBM. With reference to definite TBM, CSF TB-LAM assay had a diagnostic sensitivity of 33% and specificity of 96%. When compared to a composite reference of definite or probable TBM, the sensitivity was 24% and specificity was 95%. There were two false positive tests with TB-LAM (3+ grade). In-hospital mortality in CSF TB-LAM positive patients was 17% compared to 0% in those with definite TBM by Xpert MTB/Rif Ultra but negative LAM. Conclusions: Lumbar CSF TB-LAM has a poor performance in diagnosing TBM. Both urine TB-LAM and Xpert Ultra should be further investigated in the diagnosis of TBM.


Introduction
In many human immunodeficiency virus (HIV) endemic countries, tuberculous meningitis (TBM) is the second most common cause of adult meningitis after cryptococcal meningitis 1 , and accounts for 1-5% of all tuberculosis (TB) cases 2 . TBM is the most severe form of TB and causes substantial morbidity and mortality in children and immunocompromised adults 3,4 . HIV infection is known to increase the risk of death in patients with TBM, as does TBM stage at the time of treatment initiation 2,5 . As is the case in cryptococcosis, high-quality nursing care is a critical component in managing TBM patients 6 .
Similarly, diagnosis of TBM is very challenging, especially in resource-limited settings where diagnosis relies on a combination of clinical, radiological and laboratory findings. The World Health Organisation (WHO) recommends Xpert MTB/RIF Ultra for the diagnosis of TBM using cerebrospinal fluid (CSF). Culture has many limitations related to turnaround time and sensitivity, and also requires considerable infrastructure and costs 7 . Therefore, the development of early point of care diagnosis for TBM is a priority. Recent studies have demonstrated that the next generation Xpert MTB/RIF Ultra is the most sensitive diagnostic test in HIV-positive adults 7 . However, Xpert MTB/RIF Ultra is not a bedside test, and thus access to same day results remain a challenge in many settings 7 .
Assays based on the detection of mycobacterial lipoarabinomannan (TB-LAM) antigen in urine have emerged as potential

Diagnostic tests
In addition to standard microbiology analysis, CSF was tested with TB-LAM (Alere, Massachusetts, USA), and the test strip interpreted as per manufacturer's instructions. Briefly, the protective foil cover was removed from each test and the strip labelled with the participant's number. Two drops (or 60μL) of CSF were added to the sample pad. The test was then read after 25 minutes under standard indoor lighting conditions. The reference card was used in interpretation of the results by holding it alongside the patient window. For positive results, purple/gray bars appeared in both the control window and the patient window of the strip. For negative results, one purple/ gray bar appeared in the control window of the strip and no bar appeared in the patient window of the strip. If there was no bar

Amendments from Version 1
Based on the two reviewers' comments to version 1, major changes were made to the conclusion and Table 1 while minor changes were made to the results section as seen in the tracked changes. We added the suggested reference in the discussion section. Table 1 was modified to present the characteristics of patients with diagnosis of definite, probable, possible and not-TBM. A sentence was added on cerebral imaging. We also added a sentence about the patients who were positive with urine LAM and those who were positive with urine Ultra. We added an explanation about the 17% of patients had unknown outcome. A limitation was added that TB culture was not used. However, we have found Ultra to be more sensitive than culture (70% versus 43% against definite/probable) for the diagnosis of TBM in our population (Bahr, Lancet ID, 2018). In the conclusion section, we have mentioned the Fuji LAM that is a new LAM POC test that has higher sensitivity in urine than the determine LAM POC and should also be evaluated for diagnosis of TB meningitis both in urine and CSF. Lastly, based on the second reviewer's comments, we would like to say that a cellular CSF is common in this population with advanced HIV disease even in the setting of a confirmed TB or CM infection (Cresswell, Int J Infect Dis 2018). The not-TBM group scored <6 points on the uniform case definition (n=16) or had a confirmed alternative diagnosis (n=10 with CM).

REVISED
in the control window of the strip, the result was considered invalid and the test repeated. The strips were retained and cross checked by a second researcher to corroborate the finding.
CSF was also tested with Xpert MTB/Rif Ultra (Cepheid).
Briefly, 2ml of sample reagent was added to 1ml of whole CSF and then left to stand at room temperature for 15 minutes. Then, 2ml of the sample mixture was transferred into the Xpert MTB/Rif Ultra cartridge and loaded into the Xpert machine.
The test was run for 90 minutes and results from the assay indicate whether or not Mycobacteria TB (MTB) was detected in the sample. If MTB was detected, the results also stated whether resistance to rifampin was detected. With respect to the reference standard of definite TBM (positive CSF Xpert TB/Rif Ultra), the CSF TB-LAM assay had a sensitivity of 33% (4/12), specificity of 96% (45/47), positive predictive value (PPV) of 67% (4/6), and negative predictive value (NPV) of 85% (45/53). When compared to a composite reference of definite/probable TBM, the TB-LAM assay had a sensitivity of 24% (4/17), specificity of 95% (40/42), PPV of 67% (4/6), NPV of 76% (40/53) ( Table 2). There were two false positive tests with TB-LAM (3+ grade), without any CSF pleocytosis, normal protein, normal glucose, negative cryptococcal antigen, and normal CSF opening pressure. One patient was discharged alive without TB therapy. The second patient had a headache for 60 days at presentation, but they were lost to follow up (i.e. self-discharged) without an etiologic diagnosis. In-hospital mortality in CSF TB-LAM positive patients was 17% (1/6) compared to 0% (0/8) in those with definite TBM by Xpert MTB/Rif Ultra but negative LAM. About 17% of patients had unknown outcome. This was because the study population included patients screened for a clinical trial but only a minority were subsequently enrolled into the trial. We endeavoured to follow screen failures through to hospital discharge but this was not possible in all cases.

Conclusion
In conclusion, a rapid CSF point of care test for TBM is needed; however, this study demonstrated a poor diagnostic performance of the existing Alere TB-LAM on CSF among HIV-associated tuberculous meningitis. Our results corroborate the findings of a recent Zambian study which demonstrated 22% sensitivity for CSF LAM against a reference standard of TB culture 18 . While the relatively modest sample size is a limitation, a larger sample size is unlikely to fundamentally alter the findings of sensitivity. One explanation could be that TB-LAM is likely not be found in sufficient quantities in lumbar CSF. TB culture was not used, which is also a limitation of the accuracy analysis. However, Xpert Ultra has a sensitivity that is greater than culture in our setting 7 . The novel Fujifilm SILVAMP TB-LAM (FujiLAM) assay has been shown to have higher sensitivity in urine than the Alere TB-LAM and Data presented are the percentage, numerator/denominator, and 95% confidence intervals (CI). Test Accuracy = overall probability that a patient will be correctly classified. PPV = Positive predictive value, NPV = negative predictive value, TBM = tuberculous meningitis.  warrants evaluation for diagnosis of TB meningitis both in urine and CSF 19 . allow for better specificity. I note that some CSF's were acellular, what were these diagnoses?

Data availability
Please provide detail (a table with confirmatory findings for TB such as CXR, abdominal ultrasound, scan findings, CSF findings etc.) regarding the aspects considered to arrive at a diagnosis of probable and possible TBM. A similar consideration for possible TBM. This would add clarity on the reliability of the test outcomes.

Is the work clearly and accurately presented and does it cite the current literature? Yes
Is the study design appropriate and is the work technically sound? Yes

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate? Yes sensitivity in urine than the determine LAM POC and should also be evaluated for diagnosis of TB meningitis both in urine and CSF.
We have now added this suggested reference in the discussion section.

Response:
Comment: In Results, I would suggest the authors to present the characteristics of patients with diagnosis of probable TB using the uniform case definitions. Table 1 has been modified to present the characteristics of patients with diagnosis of Response: definite, probable, possible and not-TBM.
Comment: I am surprised by the low proportion of probable TB as compared to definite TB. This is likely to be because we used Ultra and not culture or Xpert as the reference Response: standard. We have found Ultra to be better at confirming TBM in our population (Bahr, Lancet ID, 2018) Comment: Were cerebral imaging criteria used?
: Only about 10% (n/N) of the participants had cerebral imaging done as the CT scanner Response was dysfunctional for part of the study period.
Comment: It would be also interesting to know the proportion of patients that fit the score of possible TB meningitis using the uniform case definitions.
These have been added in Table 1.

Response:
Comment: It would be interesting to know the proportion of patients with TB positive results from another specimen than CSF, which is also an important criteria for diagnosis of probable TB meningitis.
25% of the patients were positive with urine LAM while 20% were positive with urine Response: Ultra.
Comment: One option could be to present the patients' characteristics by definite TB, probable TB and others in Table 1.
: Thank you for this suggestion. This has been done in the modified Table 1. Response Comment: How do the authors explain that 17% of patients had unknown outcome of death or alive at discharge? It is quite high in a context of a nested study in a clinical trial.
This study population included patients screened for a clinical trial but only a minority Response: were subsequently enrolled into the trial. We endeavoured to follow screen failures through to hospital discharge but this was not possible in all cases.
Comment: TB culture was not used, which is a limitation of the accuracy analysis. However, Xpert Ultra has a sensitivity that is very close to culture. This could be mentioned as a limitation.
We have found Ultra to be more sensitive than culture (70% versus 43% against Response: definite/probable) for the diagnosis of TBM in our population (Bahr, Lancet ID, 2018). However we agree addition to culture would have been positive so this has been added to the limitations.
Comment: It would be also very interesting to have the results of the urine LAM if used. In the study by Siddiqui . the urine LAM had higher sensitivity than the CSF LAM in patients with et al presumptive TB meningitis.
25% of the patients were positive with urine TB-LAM LFA (Alere). The details of the Response: urine TB diagnostics are being analysed currently and will be presented separately.
Comment: In the conclusion, the authors could mention the Fuji LAM that is a new LAM POC test 1 Comment: In the conclusion, the authors could mention the Fuji LAM that is a new LAM POC test that has higher sensitivity in urine than the determine LAM POC and should also be evaluated for diagnosis of TB meningitis both in urine and CSF.
Thank you for this suggestion. This has been added in the conclusion.

Response:
No competing interests were disclosed. Competing Interests: