Point of care testing evaluation of lateral flow immunoassay for diagnosis of Cryptococcus Meningitis in Kenya, 2017

Objectives: The objective of this study was to evaluate the performance of lateral flow immunoassay (LFA) against latex agglutination (LA), India ink and culture in point-of-care diagnosis of CM. We conducted a cross-sectional study among patients with suspected CM at Mbagathi Hospital, Nairobi, April-July 2017. Results: Of 124 capillary blood and serum and 99 cerebrospinal fluid (CSF) samples, the agreement between LFA and LA on serum was 94.4%, kappa (0.88), sensitivity (100%) and specificity (91%). LFA and LA on CSF, was 97.9%, kappa (0.96), sensitivity (100%) and specificity (96%). LFA and India ink was 96.9%, kappa (0.94), sensitivity (100%) and specificity (94.1%). On CSF culture, the agreement was 72.7%, kappa (0.43), sensitivity (100%) and specificity (64%). The agreement of LFA on capillary blood, serum and CSF was 100% with kappa (1.00), sensitivity and specificity of 100%


Introduction
Cryptococcus meningitis (CM) is a life-threatening opportunistic infection among HIV infected persons (1). In Africa, CM is the second leading cause of death in HIV-infected persons (2), with a case fatality rate (CFR) of up to 38% among outpatients and 81-100% among inpatients (3). In Kenya, up to 33% of people with AIDS develop CM (4). CM diagnosis usually occurs when meningitis is at an advanced stage and treatment is less effective (1,5,6).
Culture is the gold standard diagnostic method for CM, but it has poor sensitivity, requires approximately 100µl of cerebral spinal fluid (CSF), technical expertise, and laboratory infrastructure (7,8). Microscopy requires laboratory infrastructure, and latex agglutination (LA) has sensitivity and specificity of >99% and is less labour intensive than culture but also requires technical expertise and laboratory infrastructure. Culture and LA are not available in resource constrained settings, thus limiting their clinical utility (9).
Lateral flow immunoassay (LFA) is a point-of-care (POC) test (9)(10)(11). LFA is a qualitative test to detect capsular polysaccharide antigens of Cryptococcus species complex (Cryptococcus neoformans and Cryptococcus gattii).. LFA can use whole blood, serum, or CSF, is room temperature stable, has a rapid turnaround time of <15 minutes, is simple to perform, and can be interpreted by personnel with minimal training (9).
Most studies evaluating LFA are focused on use of serum and CSF (7,8,(12)(13)(14). There are few data on use of LFA on capillary blood (15) or evaluation of LFA in Kenya. This study aimed to determine the agreement of test results from LFA on capillary blood, serum and CSF with those from LA on serum, CSF, India ink microscopy and culture on CSF.

Study design
We conducted a hospital-based cross-sectional study from April to July 2017.

Study site and population
The study was conducted at Mbagathi Hospital, a referral facility located in Nairobi County. The evaluation targeted patient ≥18 years scheduled for lumbar puncture (LP) and routine blood sample collection for CM diagnosis.
Inclusion criteria LP requested by health care provider, availability of remnant serum and CSF (≥ 500µl) after routine Cryptococcus LA or culture was performed on patients ≥18 years with the ability and willingness to provide informed consent.

Exclusion criteria
Patients involuntarily incarcerated in the hospital for psychiatric or physical illness, any client/client with a guardian who was deemed mentally unstable or unable to provide informed consent. Clients on any antifungal treatment and clients who were having repeat LPs.
Sample size assumptions and calculation The sample size was 125 participants assuming the expected proportion of agreement between LFA and other methodologies (p) was 97.7 % and the precision (P) of 3%. The sample size was calculated using Fisher's formulae (16).

Sampling methods
We reviewed records for 6 months at Mbagathi Hospital and obtained an average of 95 cases of suspected CM in a month. To achieve a sample size of 125 patients using the estimated sampling frame of 285 within a period of 3 months, every second HIV patient ≥18 years suspected of CM and scheduled for routine LP and blood collection for CM diagnosis was enrolled after giving written consent to obtain an extra routine capillary blood and use of the remnant CSF and serum for evaluation of LFA.

Data collection
The attending laboratory technologist collected blood samples from patients as part of routine testing requested by the clinician. Serum was centrifuged and separated for LA and LFA assays. CSF samples were collected by LP and centrifuged. The supernatant was used for LFA and LA assays and the pellet for culture. Sera were tested by LA, and CSF (where available) was tested by India ink microscopy and culture for clinical management of the patient. Leftover sera and CSF samples were used for the laboratory evaluation. A minimum of 500µl of the remaining sample was aliquoted into 1.8ml cryogenic vials. The samples were stored at 4°C for a maximum of 72h or at -20°C awaiting transportation to the Central Microbiology Reference Laboratory (CMRL). Additionally, a non-routine finger prick capillary blood sample was requested from all enrolled patients. Using standardized lancets and micro capillary tubes, 1-2 drops (~50 µl) of blood was transferred into micro centrifuge tubes containing LFA specimen diluent. The test was performed at the sample collection site per manufacturer instructions. Sera and CSF sample processing were done at CMRL per laboratory standard operating procedures and manufacturer instructions ( Figure 1).

Data management and analysis
Data on the test type and test results in different sample types were entered and cleaned in MS-Excel version 2013. Statistics were calculated using GraphPad QuickCalcs (GraphPad Software, Inc., La Jolla, CA) with categorical data analysis to assess sensitivity, specificity, predictive values, confidence intervals (CIs) of proportion, overall percent agreement, and kappa (k) coefficients of India ink, LA, LFA and culture in sera, CSF and capillary blood.
Interpretation of k was per standard guidelines (17).
Comparison of culture to LA using CSF, the sensitivity and specificity were 95. Comparison of LFA on capillary blood to LFA on sera, the sensitivity and specificity were100%, PPV and NPV at 100% with a total agreement of 100% (124) and a kappa-value of 1.00 (95% CI 1.00-1.00) ( Table 1). LFA on capillary blood was compared to LFA on CSF, the sensitivity, specificity and predictive values were all 100% with a total agreement of 100% (99) and a kappa-value of 1.00 (95% CI 1.00-1.00) (Table 1a).

Conclusion
Our results show high agreement between LFA, LA and India ink in different samples and a perfect agreement between LFA in different samples. The high agreement shows that LFA is a reliable POC diagnostic test. The results on individual tests show that there was almost perfect agreement between LFA and LA on CSF and serum. The test demonstrated high level of sensitivity and specificity of LFA compared to LA on sera and CSF. These findings are consistent with similar studies conducted in South Africa and USA that show high sensitivity using CSF and serum (12,18). Comparable results were reported in a study on multisite validation of cryptococcal antigen lateral flow assay in Uganda and South Africa (7). The strong agreement between the LFA and LA tests is an indicator that LFA test on whole blood, CSF and serum is as good as LA test on sera and CSF.
The findings from comparison of LFA to India ink microscopy using CSF demonstrated high sensitivity, specificity, and predictive values. This is in contrast to the findings from the expert opinion and other studies that documented lower sensitivity and NPV for CSF microscopy against LFA (7,9). The India ink microscopy requires laboratory infrastructure, dependent on fungal concentration and is highly operator dependent rather than the test performance.
On comparison of LFA to culture using CSF, there was high sensitivity, low specificity, and moderate agreement with a weak kappa value. The findings were consistent with other studies that documented high sensitivity and low specificity (14,19). The findings on high sensitivity, low specificity and a weak kappa value were similarly demonstrated when CSF culture was compared to LA and India ink using CSF. However, other studies documented low sensitivity in CSF culture when compared against other diagnostic tests (7-9).
LFA on capillary blood was compared with LFA on serum and CSF. The LFA results on capillary blood had an ideal concordance with LFA serum and CSF results. LFA had a very high positive and negative predictive values both on serum and CSF, a characteristic that makes it good for an accurate diagnosis of cryptococcal meningitis. The high sensitivity and specificity of the test and its ability to be easily performed at the bedside and giving accurate results rapidly allows for prompt and timely initiation of treatment (9). The findings were comparable with a similar study on evaluation of LFA using serum, CSF and capillary blood in Uganda (15,18,20).

Recommendation
The evidence of the perfect agreement between LFA on capillary blood, serum and CSF, high sensitivity and specificity, ease of performance, along with rapid results may indicate LFA using capillary blood POC test as the method of choice for CM diagnosis. Although LFA meets World Health Organization assured criteria for POC diagnostic tests in resourcelimited settings, we recommend use of LFA test as a POC test in resource limited settings for the diagnosis of CM.

Limitation of the study
The limitations of this study include participants not yielding CSF sample due to dry taps, thus reducing the CSF samples that were analysed. The difference on CSF samples analysed had no major implications to the overall evaluation since there were more than one sample type used in the evaluation. Capillary blood could only be used on LFA test thus, there was no uniform use of the sample type across other testing procedures.

Availability of data and material
All data related to this study are available upon request.
Ethics approval and consent to participate

Funding
Financial support for this research was provided by Africa Field Epidemiology Network (AFENET). AFENET had no role in design of the study and collection, analysis, and interpretation of data and in writing the manuscript.

Consent for publication
Not applicable.

Competing interests
The authors report no conflicts of interest.   Figure 1 Evaluation of LFA Algorithm