PIMA™ point-of-care testing for CD4 counts in predicting antiretroviral initiation in HIV-infected individuals in KwaZulu-Natal, Durban, South Africa

Introduction Limited information is available on the usefulness of the PIMA™ analyser in predicting antiretroviral treatment eligibility and outcome in a primary healthcare clinic setting in disadvantaged communities in KwaZulu-Natal, South Africa. Materials and methods The study was conducted under the eThekwini Health Unit, Durban, KwaZulu-Natal. Comparison of the enumeration of CD4+ T-cells in 268 patients using the PIMA™ analyser and the predicate National Health Laboratory Services (NHLS) was undertaken during January to July 2013. Bland-Altman analysis to calculate bias and limits of agreement, precision and levels of clinical misclassification at various CD4+ T-cell count thresholds was performed. Results There was high precision of the PIMA™ control bead cartridges with low and normal CD4+ T-cell counts using three different PIMA™ analysers (%CV < 5). Under World Health Organization (WHO) guidelines (≤ 500 cells/mm3), the sensitivity of the PIMA™ analyser was 94%, specificity 78% and positive predictive value (PPV) 95%. There were 24 (9%) misclassifications, of which 13 were false-negative in whom the mean bias was 149 CD4+ T-cells/mm3. Most (87%) patients returned for their CD4 test result but only 67% (110/164) of those eligible (≤ 350 cells/mm3) were initiated on antiretroviral therapy (ART) with a time to treatment of 49 days (interquartile range [IQR], 42–64 days). Conclusion There was adequate agreement between PIMA™ analyser and predicate NHLS CD4+ T-cell count enumeration (≤ 500 cells/mm3) in adult HIV-positive individuals. The high PPV, sensitivity and acceptable specificity of the PIMA™ analyser technology lend it as a reliable tool in predicting eligibility and rapid linkage to care in ART programmes.


Introduction
Poor rates of linkage to care for those with low CD4+ T-cell counts, eligible for antiretroviral therapy (ART), have been reported in several African cohort studies. 1,2,3,4,5 Several attrition steps exist in the continuum of care pathway: patients lost to care between testing HIV-positive and going for a CD4 test 6 ; CD4 test result not available and/or lost 7,8 ; patient not returning for their CD4 test result; and lack of uptake of care from eligibility to initiation of ART even in those who return for test results. 9 These challenges may be overcome by point-of-care (POC) testing 10 , resulting in less attrition over time. 2,4 It has been suggested that POC CD4 testing in those who do not return for their results would potentially increase enrolment pre-ART. 9 POC CD4 testing was shown to modestly increase linkage to care and reduce pre-treatment loss to follow-up in fixed and mobile clinics. 2,11,12,13 Factors contributing to pre-treatment loss to follow-up have been previously documented. 8,14,15,16,17 The Alere PIMA TM POC has been evaluated against the 'gold standard' flow cytometry platforms, for example, Beckman Coulter using panleucogating (PLG) 18 ; BD FACS count 19,20,21 ; PARTEC Cytoflow TM19,20 ; Guava and BD FACS Calibur 20,21,22,23,24 for the enumeration of CD4+ T-cells in HIV-1-infected adults and in HIV-1-infected pregnant women. 22 This study assessed the accuracy, sensitivity and specificity of the Alere PIMA TM POC analyser in CD4+ T-cell count enumeration compared to the predicate South African National Health Laboratory Services (NHLS) flow cytometry test (Beckman Coulter) and its potential operational role as a predictor of ART eligibility in a primary healthcare clinic (PHC) in Durban, South Africa.
Introduction: Limited information is available on the usefulness of the PIMA TM analyser in predicting antiretroviral treatment eligibility and outcome in a primary healthcare clinic setting in disadvantaged communities in KwaZulu-Natal, South Africa.

Materials and methods
The study occurred at Lancers Road PHC, a facility under the eThekwini Health Unit, situated in the centre of the convergence of the taxi rank in the city centre of Durban. This PHC offers HIV Counselling and Testing (HCT)approximately 900 per month to walk-in patients who receive pre-and post-test counselling and CD4 testing for the staging of HIV-1-infected disease to determine eligibility for ART. Patients are advised to return after 7 days for their CD4 results. As per the SA HIV and AIDS guidelines 25 operating at the time of this study, patients with a CD4+ T-cell count ≤ 350 cells/mm 3 , upon their return, were medically assessed, and education and counselling undertaken prior to ART initiation. Those ineligible for ART, viz CD4+ count > 350 cells/mm 3 , were counselled to return 6 monthly for CD4+ T-cell count testing and for further medical assessment. Eligible patients, who did not return for results, were contacted telephonically to ascertain whether they had been initiated on ART elsewhere, and if not, they were encouraged to return for further care.

Testing of venous blood samples
Routine CD4+ T-cell enumeration is conducted at the NHLS one day after the blood draw via Beckman Coulter flow cytometry using PLG methodology, the standard of care in this setting as described previously. 26 During January 2013 to July 2013, in 268 patients, an extra 2 mL of venous blood was drawn from the same blood draw as the routine NHLS test into another ethylenediaminetetraacetic acid (EDTA) tube for the comparison of the enumeration of CD4+ T-cells using the Alere PIMA TM technology (Alere Health Care, Waltham, Massachusetts). PIMA TM POC CD4+ T-cell enumeration was conducted by a laboratory technician who pipette-filled the PIMA TM cartridges. Three PIMA TM analysers were used in this study. CD4+ T-cell count enumeration was performed in a subset of 100 samples using the FACS Calibur.

Quality control and/or precision of PIMA TM analysers
Quality control and routine PIMA TM analyser maintenance were performed daily as per manufacturer's guidelines: one control has low CD4+ T-cell counts (115 cells/mm 3 -235 cells/mm 3 ) and the other has normal CD4+ T-cell counts (719 cells/mm 3 -1355 cells/mm 3 ). Daily quality control was conducted on all 3 analysers for the first 10 measurements when a new cartridge was used and over a period of 165 days (23 January -25 March 2014). Accuracy and precision of the NHLS PLG testing was established in the NHLS laboratories by daily monitoring of instrument stability (Flow check TM, Beckman Coulter Miami, FL) and system performance verification using normal (394 cells/mm 3 -754 cells/mm 3 ) and low (62 cells/mm 3 -206 cells/mm 3 ) Immunotrol TM controls (Beckman Coulter, Miami, FL). The Addington NHLS laboratory participates in the NHLS proficiency testing panels and is accredited by the South African National Accreditation System. 27

Reproducibility of CD4+ T-cell enumeration across flow cytometry instruments
Comparisons of CD4+ T-cell enumeration was undertaken between flow cytometry instruments (PIMA TM POC analysers and the predicate NHLS) on 268 blood samples. Due to transport logistics, the NHLS laboratory performs testing the day after the blood draw. Therefore, a subset of 100 blood samples were tested by the PIMA TM analyser, FACS Calibur and the NHLS to ensure that differences observed between the PIMA TM analyser versus NHLS were not due to CD4 testing performed on the next day in the NHLS laboratory. CD4+ T-cell enumeration using the FACS Calibur reference method 28 was undertaken on the same blood sample tube as the PIMA TM POC analyser at the Medical Research Council Central laboratory, which participates in the United Kingdom National External Quality Assessment Scheme (UK NEQAS) quality assessment programme.

Predictions of benefit of PIMA TM POC CD4 test results for ART eligibility and linkage to care
Prediction of the benefits of the PIMA TM POC CD4 testing in terms of ART eligibility and decision making was undertaken. Additionally, an assessment was undertaken to determine whether HIV-infected individuals return for their CD4+ test result and how many are lost to follow-up between ART eligibility and initiation.
The protocol was approved by the Biomedical Research Ethics Committee, University of KwaZulu-Natal (BE 212/11) and the eThekwini Research Ethics committee (28 November 2011). Written informed consent was obtained from patients > 18 years of age enrolled in the study.

Statistical analysis
It was determined that a sample size of 254 HIV-positive patients would be required to detect a difference of 15 cells/mm 3 between the results of the PIMA TM POC analyser and the conventional test with 95% probability and 80% power assuming the standard deviation of difference in means is 85. In order to allow for potential problems with samples, the sample size was increased by 14 patients giving a sample size of 268.

Statistical methods
Pairwise comparison of the PIMA TM analysers was conducted using t-tests. To assess the precision of the control cartridge within each of the three PIMA TM analysers, the %CV was calculated for the 10 observations (intra-day reproducibility) and over a period of 165 days (inter-day reproducibility) at low and normal beads.
The percentage similarity (% SIM) model, Bland-Altman (BA) plots, limits of agreement (LOA) and Lin's concordance correlation coefficient were used to assess agreement between PIMA TM analysers, FACS Calibur and NHLS. 29 To assess the diagnostic accuracy of CD4+ T-cell counts by the PIMA TM POC analysers in identifying ART eligibility, sensitivity, specificity, false-negative (FN) and false-positive (FP) rates, positive predictive value (PPV) and negative predictive value (NPV) were computed for the ART initiation thresholds of ≤ 200 cells/mm 3 , ≤ 350 cells/mm 3 and ≤ 500 cells/mm 3 CD4+ T-cells. All analyses were performed using STATA (Statacorp, College Station, TX, USA) statistical version 13.

Reproducibility of results of PIMA TM machines used in this study
There was high reproducibility and instrument precision (%CVs < 5%) within PIMA TM analysers 1, 2, 3 of the control cartridges over a replicate set of 10 bead analyses and over time (n = 165 days; 23 January -25 March 2014). The bead quality control (QC) count for low and normal bead cartridges showed median %CV results for the 10 same-day observations of 2.13%, 1.28%, and 1.41% and 0.86%, 1.36%, and 0.96% for analysers 1, 2, and 3, respectively. Bead QC counts for low and normal bead cartridges showed median %CV results over the 165 days of 1.75%, 1.70%, and 1.86% and 1.14%, 1.67%, and 1.30% for PIMA TM analysers 1, 2, and 3, respectively.
The majority (218/268) of HIV-1-positive individuals undergoing CD4+ T-cell count testing were women of whom 25% were 25-29 years, whereas the majority of the men were older than 30 years (Table 1). There was no significant difference in the median CD4+ T-cell count between men and women performed by the NHLS versus the PIMA TM POC analyser, although the median CD4+ T-cell count was higher in the latter. According to the NHLS versus PIMA TM POC, 81% versus 80% of HIV-positive individuals were eligible for ART initiation (≤ 500 cells/mm 3 ), of whom 82% versus 84% were males and 81% versus 79% were females, respectively.
In a subset of 100 samples, the highest agreement was observed between PIMA TM analysers and FACS Calibur as evidenced by smaller mean bias of 7.52 and narrower BA limits of agreement from -111 to 126 and a correlation of 0.97 (Table 2). Wider BA limits of agreement (from -216 to 176 mean bias -20.3) were observed between the FACS Calibur versus NHLS with a correlation of 0.92 compared to PIMA TM analysers versus NHLS (BA limits of agreement from -226 to 200 mean bias -12.78) with a correlation of 0.90.
Under previous SA ART guidelines of ≤ 200 cells/mm 3 and ≤ 350 cells/mm 3 , the PIMA TM POC analysers displayed a sensitivity and specificity of 73.5%/98.4% and 83.5%/92.3%, respectively (Table 4). Under the current SA guidelines of ≤ 500 CD4+ T-cells/mm 3 , a high sensitivity of 94% and PPV of 95% was observed at the sacrifice of lower specificity of 78%. In the 13 FNs with ≤ 500 cells/mm 3 , the mean bias was 149 CD4+ T-cells/mm 3 .
As the study was conducted during 2013, linkage to care data is presented according to the NHLS laboratory CD4 test result of ≤ 350 cells/mm 3 , 25 164/268 (61%) of patients were eligible for ART on the day of HCT compared to 145/268

Discussion
Conventional flow cytometry to determine CD4 counts usually requires that samples be sent to a central laboratory, which may be off-site. Although the turn-around time for a CD4 test result by the NHLS is 24-72 hours, HIV-1-infected patients are counselled to return to the PHC within 1 week for receipt of these results. POC technologies can reduce these delays resulting in rapid linkage to care. This study demonstrated a high PPV and sensitivity and acceptable specificity in predicting ART eligibility (≤ 500 cells/mm 3 ) using the PIMA TM POC analyser as compared to the NHLS CD4 test.
The majority of HIV-1-positive individuals undergoing CD4 testing were women, of whom 25% were 25-29 years old, whereas the majority of men were older than 30 years of age. There were no significant differences in the median CD4+ T-cell count in men versus women performed by the NHLS versus the PIMA TM POC analyser, although the median count was higher in the latter. Overall, according to NHLS versus PIMA TM POC, 81% versus 80% of individuals were eligible for ART initiation (≤ 500 cells/mm 3 ), of whom 82% versus 84% were men and 81% versus 79% were women, respectively.
There was high reproducibility in all three PIMA TM POC analysers using normal and low beads with coefficient of variation < 5% over time (10 and 165 days). The PIMA TM POC analyser slightly overestimates NHLS flow cytometry in CD4+ T-cell enumeration in this study, which corroborates most studies using capillary or venous blood. 26,30,31,32 This overestimation is minimal (mean bias 17 cells/mm 3 ) and is not clinically significant. Differences have been reported on conventional CD4 testing platforms between the BD FACS count versus the BD FACS Calibur 23 where the mean bias between the two platforms was -76 cells/mm 3 (95% CI LOA -316.0-163.0).
The adequate correlation between the PIMA TM POC analyser and FACS Calibur (0.97) corroborates similar findings in another study. 21 Although a correlation of > 0.90 was observed between the three platforms, these differences are due to variability of instrument settings, antibodies and fluorochromes used, gating strategies and sample volume input.

Measure of agreement PIMA analysers -NHLS † PIMA analysers -FACS Calibur † FACS Calibur -NHLS †
Mean  where all those eligible for treatment will be initiated but it will come at a cost of low specificity, whereby individuals not needing treatment will be commenced on ART. However, in light of ART-lowering viral loads and reducing horizontal transmission, 36,37 this downside is minimised. A recent study has demonstrated that as household ART coverage is increased, there is a decrease in HIV acquisition. 38 The agreement in these data between the PIMA TM POC analyser and NHLS laboratorybased flow cytometry appears to decline with increasing CD4+ T-cell count ≥ 500 cells/mm 3 . This is not of concern as these HIV-1-infected individuals are ineligible for ART under current guidelines.
From the operational perspective in the use of the PIMA TM POC analyser, similar to other studies using venous or capillary blood, 18,20,21,22,33 we also experienced reading errors (8%) mostly because of movement and vibration. 34 The 'operator' used in our study was a trained laboratory technician compared to health professionals, for example, nurse or counsellor. Several studies have reported that the PIMA TM POC is interchangeable with conventional platforms, 18 A recent study reported that providing same-day POC CD4 testing that is not rapid has no benefit in health outcomes. 9 As suggested by others, 11,18,40,41 we agree that using existing infrastructure and based on demand, the integration of a PHC POC mini-laboratory run by dedicated personnel (laboratory technician) is possible, offering tests for staging and pathology that assess ART eligibility. 42,43 However, as suggested in a recent systematic review, 44 this needs to be supported by streamlining services through minimising patient clinic visits, 22 addressing psychosocial issues and barriers to healthcare, 45 optimising the opportunity for patient empowerment through counselling and peer support, 46 emphasising the importance of starting and adhering to ART if eligible, 6 positive healthseeking behaviours and encouragement for patient ownership of their health. A family-centred model of integrated healthcare incorporating most of the abovementioned health system changes has previously been shown, in a similar population, to yield high adherence (94%) and retention in the care and management of HIV-1-positive individuals. 47,48 In this study, similar reasons for not linking into care were given as found previously 45 ; of those eligible for ART who did not access treatment (33%), the reasons given upon telephonic communication were economic (no money to cover transport costs), social (too busy to come to the clinic), structural (cannot take time off work) and emotional (were not ready to take ART and they were still

Conclusion
In summary, the overall agreement between PIMA TM POC analyser and NHLS CD4+ T-cell count enumeration in adult HIV-1-positive individuals was acceptable with clinically insignificant mean bias. Together with high PPV and sensitivity and acceptable specificity, the PIMA TM POC CD4 test has the potential role for CD4+ T-cell enumeration in PHC settings and lends itself to be an excellent facilitator in rapid linkage to care in ART programmes, particularly that it has been demonstrated in simulated cohort models of HIV-1-infected adults and pregnant women, to result in not only better clinical outcomes but also to cost savings in the long term. 49,50 Even in the era of 'test and treat', 51 PIMA TM POC CD4 testing would facilitate the fast tracking of patients with low CD4+ T-cell counts (< 200 cells/mm 3 ) for the administration of cotrimoxazole prophylaxis as well as in screening for cryptococcal infection in patients with < 100 cells/mm 3 . The operational role of the PIMA TM POC CD4 test in provision of immediate CD4+ T-cell count results combined with integrated health system changes and interventions such as mobile phone technology and provision of incentives need to be evaluated in a variety of settings across the HIV cascade, to determine its implementation effectiveness in linkage to care, time to ART initiation and retention in HIV care.