Sensitive Blood-Based Detection of HIV-1 and Mycobacterium tuberculosis Peptides for Disease Diagnosis by Immuno-Affinity Liquid Chromatography–Tandem Mass Spectrometry: A Method Development and Proof-of-Concept Study

Abstract

BACKGROUND:

Novel approaches that allow early diagnosis and treatment monitoring of both human immunodeficiency virus-1 (HIV-1) and tuberculosis disease (TB) are essential to improve patient outcomes.

METHODS:

We developed and validated an immuno-affinity liquid chromatography–tandem mass spectrometry (ILM) assay that simultaneously quantifies single peptides derived from HIV-1 p24 and Mycobacterium tuberculosis (Mtb) 10-kDa culture filtrate protein (CFP10) in trypsin-digested serum derived from cryopreserved serum archives of cohorts of adults and children with/without HIV and TB.

RESULTS:

ILM p24 and CFP10 results demonstrated good intra-laboratory precision and accuracy, with recovery values of 96.7% to 104.6% and 88.2% to 111.0%, total within-laboratory precision (CV) values of 5.68% to 13.25% and 10.36% to 14.92%, and good linearity (r² > 0.99) from 1.0 to 256.0 pmol/L and 0.016 to 16.000 pmol/L, respectively. In cohorts of adults (n = 34) and children (n = 17) with HIV and/or TB, ILM detected p24 and CFP10 demonstrated 85.7% to 88.9% and 88.9% to 100.0% diagnostic sensitivity for HIV-1 and TB, with 100% specificity for both, and detected HIV-1 infection earlier than 3 commercial p24 antigen/antibody immunoassays. Finally, p24 and CFP10 values measured in longitudinal serum samples from children with HIV-1 and TB distinguished individuals who responded to TB treatment from those who failed to respond or were untreated, and who developed TB immune reconstitution inflammatory syndrome.

CONCLUSIONS:

Simultaneous ILM evaluation of p24 and CFP10 results may allow for early TB and HIV detection and provide valuable information on treatment response to facilitate integration of TB and HIV diagnosis and management.

Introduction

People living with human immunodeficiency virus (HIV) infections (PLHIV) have a 15-to 21-fold increased risk of developing tuberculosis (TB), and TB is the leading cause of morbidity and mortality among PLHIV (1). In 2021, there were an estimated 1.6 million TB-related deaths, including 187 000 deaths in PLHIV (2). The World Health Organization (WHO) recommends regular screening for active TB during HIV care visits (3). Similarly, the U.S. Centers for Disease Control and Prevention (CDC) recommends HIV screening for all patients with TB disease or latent TB infection (4). Intensifying TB case-finding in PLHIV and providing HIV testing and counseling to patients with presumptive and diagnosed TB are necessary steps to mitigate disease burden in populations at risk for, or affected by, both diseases (5). WHO policy thus emphasizes the need to establish mechanisms that deliver integrated TB and HIV services, preferably at the same time and location (6).

Frontline sputum-based TB diagnostics (e.g., acid-fast bacilli [AFB] smear, Mycobacterium tuberculosis [Mtb] culture, and GeneXpert MTB/RIF [Xpert] nucleic acid amplification) have low sensitivity in individuals with low Mtb concentrations (paucibacillary), which often occur in the context of HIV/TB co-infection (7, 8). Sensitive PCR-based tests for HIV and Mtb exist, but are approved for use with different specimen types, reducing their utility in joint screening efforts. Novel approaches that permit sensitive and specific diagnosis of HIV/TB via multiplex serum biomarker assays would therefore address an urgent need to improve their integrated disease management.

Measuring circulating levels of pathogen-specific proteins can diagnose infection and detect changes in pathogen burden that reflect treatment response (9). Immunoassays used for this purpose can be subject to confounding factors (10, 11), including interactions with host factors, such as antibodies, which can attenuate sensitivity, and false-positive detection of related factors can reduce specificity. However, these inhibitory effects can be mitigated by using protein dissociation protocols and high-specificity antibodies, or avoided by use of immunoassays that employ liquid chromatography–tandem mass spectrometry (LC-MS/MS) to quantify biomarker peptides released from diagnostic specimens, as this disrupts protein complexes and allows detection of biomarker peptides that distinguish highly conserved protein isoforms.

Herein, we describe a blood-based immuno-affinity liquid chromatography–tandem mass spectrometry (ILM) assay that quantifies proteotypic peptides derived from the Mtb virulence factor 10-kDa culture filtrate protein (CFP10) and the HIV-1 capsid protein p24 (Fig. 1A) to simultaneously diagnose HIV-1 infection and TB disease and to monitor their real-time response to treatment.

Fig. 1.

Characteristics of the p24/CFP10 ILM assay. (A), Schematic of the assay procedure. SRM transitions (B, C) and LC retention profiles (D, E) of seven targeted p24pep and CFP10pep y-ions. Standard curves for (F) p24pep and (G) CFP10pep detected by the p24/CFP10 ILM assay and imprecision for low, intermediate, and high concentrations of (H) p24pep and (I) CFP10pep.

Materials and Methods

SERUM AND PLASMA SAMPLES

Cryopreserved plasma samples from adults with active TB disease (AFB-culture and/or GeneXpert confirmed) including individuals with HIV (n = 23), and their household contacts (asymptomatic without evidence of TB disease, n = 11) enrolled in the TB Aerobiology, Infectiousness, and Transmission (TBAIT) cohort in Kenya were selected for inclusion in this pilot evaluation if they had sufficient sample volume for analysis and segregated according to their HIV-1 infection and TB disease status. Cryopreserved serum samples from children with TB/HIV-1 (n = 13) and control children (n = 4), who had perinatal HIV-1 exposure but did not have HIV-1 infections or TB disease, were obtained from the sample archive of the International Maternal-Pediatric-Adolescent AIDS Clinical Trials (IMPAACT) P1041 study, a multi-center phase II–III randomized double-blind placebo-controlled trial that evaluated primary isoniazid (INH) prophylaxis in bacille Calmette-Guérin (BCG)-vaccinated 3-to 4-month-old children with perinatal HIV-1 exposure in southern Africa between 2004 and 2008 (12). All study participants (TBAIT) or their legal guardians (P1041) provided written informed consent in compliance with an Institutional Review Board (IRB)-approved protocol prior to study enrollment.

Five commercial HIV-1 seroconversion panels (SeraCare AccuVert 0600–0240, 0600–0244, 0600–0246, 0600–0272, and 0600–0291) containing undiluted serial plasma samples collected from 5 individuals within 17 to 50 days of initial evaluation for acute HIV infection were purchased from SeraCare Life Sciences Inc. Serial p24 and HIV-1 RNA reports for these samples were from the Abbott ARCHITECT HIV Ag/Ab Combo assay, Roche COBAS Elecsys HIV combi PT assay, Perkin Elmer HIV-1 p24 ELISA, and Roche COBAS AmpliPrep/COBAS TaqMan HIV-1 Test v2.0 and results were supplied with the samples.

HIV AND TB DIAGNOSTIC CRITERIA

All TBAIT and P1041 study participants were assessed for HIV-1 infection at enrollment. P1041 study participants with negative HIV-1 PCR assay results at enrollment were re-evaluated 3 and 15 months post-enrollment with PCR- and ELISA-based HIV-1 assays, respectively, and considered HIV-negative only if the results of both tests were negative. P1041 study participants who had samples analyzed in this study were retrospectively assigned TB classifications using the 2015 NIH criteria for pulmonary TB in children (13). Those with culture-positive results or >2 non-bacteriological forms of TB evidence were classified as TB-positive, and those with no or insufficient evidence of TB, or a confirmed alternative diagnosis, were designated as a TB-negative group for this study. TBAIT study participants were offered HIV testing as part of TB diagnosis at enrollment, and participants with negative HIV antibody tests were re-evaluated at 6 and 12 months post-enrollment. TBAIT study participants who had samples analyzed in this study were defined as having TB as confirmed by AFB-culture and/or GeneXpert sputum results or as non-TB household contacts without any TB diagnostic criteria.

PEPTIDES AND PEPTIDE-SPECIFIC MONOCLONAL ANTIBODIES

Synthetic HIV-1 p24 and Mtb CFP10 target peptides (p24pep: ETINEEAAEWDR and CFP10pep: TDAATLAQEAGNFER) and their sequence-matched, stable isotope-labeled internal standard peptides (${\hspace{1em}}^{13}{\text{C}}_6^{15}{\text{N}}_4$ C-terminal arginine label) were purchased from GenScript. Rabbit monoclonal antibodies (mAbs) raised against keyhole limpet hemocyanin-conjugated p24pep and CFP10pep peptides (GenScript) were purified from culture supernatants using their target peptides or protein A (>99% pure by SDS-PAGE), resuspended in PBS, A280-quantified with a NanoDrop spectrophotometer, and aliquoted to generate stock solutions (anti-CFP10: 0.524 mg/mL, anti-p24: 1.24 mg/mL) that were stored at −80°C until use.

SAMPLE PREPARATION, ANALYSIS AND ASSAY VALIDATION

Plasma or serum samples were mixed with lysis buffer, heat denatured, and then pH adjusted and trypsin digested overnight before incubation with magnetic beads conjugated with HIV-1 p24 or Mtb CFP10 peptide specific mAbs (Fig. 1A) and peptides eluted from these beads were then desalted on a trap column and analyzed on an Evosep One system (Evosep) coupled with a TSQ Altis mass spectrometer (ThermoFisher Scientific; see online Supplemental Material). Validation studies were performed to evaluate assay linearity, lower limit of the measuring interval (LLMI), imprecision, immunoprecipitation yield, sample carryover contamination, and matrix effects, as described in the Supplemental Material.

CRITERIA FOR P24PEP AND CFP10PEP PEAK IDENTIFICATION

Target peaks were identified using previously published acceptance criteria for peak data quality and noise reduction (14). Specifically, all peaks were required to meet all of the following criteria to be deemed target peptide signal: signal-to-noise ratio (SNR) ≥2.2, dot product (dotp) and relative dot product (rdotp) similarity scores ≥0.71 and ≥0.79, with negatively correlated ion pair percentages $(\%{{\text{ρ}}_{{\text{i}}_{\text{P}}}}^{-})$ and the sum of their significantly correlated ion pair coefficients $({\sum }_{\mathrm{s}}{{\text{ρ}}_{{\text{i}}_{\text{P}}}}^{\pm })$ having values that were <22% and ≥1.58, respectively. Samples that produced candidate p24pep or CFP10pep that failed one or more quality assurance parameters were subjected to repeated analysis if they contained sufficient volume. Samples that contained peptides that met these acceptance criteria were considered positive for these peptides.

STATISTICAL ANALYSIS

Skyline 22.2 software (MacCoss Lab Software) was used to analyze LC-MS/MS spectra against a library produced using recombinant p24 and CFP10 digests. Calculation of diagnostic sensitivity and specificity were performed with the MedCalc Diagnostic test evaluation calculator and/or STATA (version 15). All other statistical analyses were done with SPSS (version 26.0). GraphPad Prism (version 9.4.1) was used for data visualization. MS data has been deposited to the Panorama Public repository with the link https://panoramaweb.org/g9ah75.url.

Results

ILM ASSAY DEVELOPMENT

To develop the ILM assay for HIV-1 and Mtb infection, we first selected peptides derived from the HIV-1 capsid protein p24 and the Mtb virulence factor CFP10, which have been previously employed to diagnose HIV-1 infection and TB disease but have limitations when analyzed by conventional immunoassays. CFP10 is actively secreted from early infection onward and is critical for Mtb virulence and pathogenesis (15, 16) but conventional blood-based immunoassays that target intact CFP10 protein can be compromised by host antibodies, similar to the rapid p24 masking effect observed following HIV infection (17). We have previously reported that a CFP10 peptide (TDAATLAQEAGNFER; CFP10pep) is readily detected in trypsin-digested blood samples of individuals with TB, irrespective of age, sex, immune function, or the site(s) of TB (pulmonary, extrapulmonary) when analyzed by MS (10, 14, 18–20). We therefore analyzed HIV-1 p24 using a previously published peptide biomarker discovery workflow (14) to identify a target peptide for an ILM HIV-1 assay (Fig. 1A and online Supplemental Fig. 1).

Comparison of the LC-MS/MS characteristics of CFP10pep and p24pep found that the 7 major y-ion transitions of their synthetic peptides produced consistently strong signal intensities (Fig. 1B and C, and online Supplemental Table 1), although there was an approximate 17-fold difference in CFP10pep vs p24pep integrated signal intensity (online Supplemental Fig. 2), and eluted with widely separated LC gradient retention times to permit good resolution of their ion signals (Fig. 1D and E). Rabbit mAbs raised against p24pep and CFP10pep peptides also revealed similar target recoveries when employed to immunoprecipitate 1 pmol/L of p24pep and CFP10pep synthetic peptides spiked into healthy human plasma (mean recoveries of 83.1 ± 6.1% and 90.0 ± 7.4%, respectively, online Supplemental Fig. 3).

ILM ASSAY VALIDATION

An integrated and optimized p24/CFP10 ILM assay generated with these 2 mAbs demonstrated good linearity (Pearson r² values >0.99) across a wide linear range for both p24 (1 to 256 pmol/L) and CFP10 (0.016 to 16 pmol/L) (Fig. 1F and G). No positive signal was detected in the 0.5 pM p24 and 0.008 pM CFP10 standards, as these samples consistently failed to meet the peak similarity criteria for positive signal (online Supplemental Table 2). An LLMI evaluation was thus performed using the lowest consistently detected standard curve samples for p24 and CFP10 (1.0 and 0.016 pmol/L). Both these standards revealed CVs <20% (13.3% and 14.9%) and total errors of ±15% (−4.4% and −12.0%) to meet the acceptance criteria for LLMI values. Intra-assay CVs for 3 p24 and CFP10 concentrations within the linear range of this assay varied between 0.70% and 19.64%, and 1.24% and 15.09%, respectively, with mean inter-assay p24 and CFP10 CVs ranging between 5.68% and 13.25%, and 10.36% and 14.92%, respectively, across these concentrations (Fig. 1H and I, and online Supplemental Table 3). Recoveries (%) for p24 and CFP10 spiked into aliquots of a pooled sample generated from 10 HIV/TB patient samples varied from 96.7% to 104.2% and 88.2% to 111.0%, respectively, with calculated biases ranging from −11.83% to 11.00% (Table 1).

Table 1.

Recoveries of p24 and CFP10 detected by the ILM assay.

	Starting concentration, pmol/La	Spike-in concentration, pmol/La	Observed concentration, pmol/L	Average recovery (%)	Bias (%)
p24	9.10	9.00	18.09
			19.45	96.74	−3.26
			15.88
		25.00	36.10
			31.76	101.88	1.88
			35.85
		50.00	61.94
			66.23	104.55	4.55
			55.95
CFP10	0.064	0.100	0.220
			0.169	111.00	11.00
			0.136
		0.200	0.262
			0.242	88.17	−11.83
			0.217
		0.300	0.384
			0.349	104.67	4.67
			0.401

^aPooled serum aliquots generated from equal volumes of serum from 10 individuals with HIV and TB were spiked with the indicated concentrations of recombinant p24 and CFP10 protein.

No substantial carryover of p24pep and CFP10pep signal was detected in negative control healthy serum samples analyzed directly after high concentration (256 pmol/L CFP10 and 16 pmol/L p24) positive control samples (online Supplemental Table 4). Nor was significant interference detected when plasma samples containing low p24 and CFP10 concentrations were spiked with 10 mg/mL (11.29 mmol/L) triglycerides or 0.4 mg/mL (684 μmol/L) bilirubin. However, addition of 10 mg/mL hemoglobin or 100 mg/mL albumin decreased measured p24 (7.33% and 12.42%) and CFP10 (11.56% and 12.75%) concentrations (online Supplemental Table 5). We next supplemented healthy human serum samples spiked with low-concentration CFP10 and p24 with varying amounts of albumin and hemoglobin to further investigate their effect on biomarker signal. Both interferents caused progressive recovery losses, although hemoglobin had a more pronounced effect than albumin (online Supplemental Fig. 4). However, these losses did not directly correlate with total or added protein, and increasing the amount of trypsin did not improve target recovery (online Supplemental Fig. 5), indicating that these were not due to incomplete digestion of CFP10 and p24 due to excess protein and thus could not be compensated by the use of additional trypsin. Recovery variance was greater for p24 than CFP10, as p24 was analyzed near its LLMI, unlike CFP10.

MEASUREMENT OF SERUM/PLASMA P24 AND CFP10 LEVELS IN PARTICIPANTS WITH HIV, TB, AND HIV/TB

We next evaluated the performance of this assay in cryopreserved serum or plasma samples obtained from the biorepositories of an adult TB and household contact study (TBAIT, n = 34) and a TB prevention trial in children (P1041, n = 17) that included samples from individuals diagnosed with HIV-1 and/or TB, or no TB/HIV-1 (Table 2). The ILM p24 assay detected positive signal in 12 of 14 adults and 8 of 9 children with HIV at baseline, while the ILM CFP10 assay detected positive signal in all 23 adults and 8 of 9 children with TB. ILM assay results had similar diagnostic sensitivity for HIV-1 and TB in the adult and child cohorts (85.7 [95% CI, 57.2–98.2] % vs 88.9 [95% CI, 51.8–99.7] % for HIV and 100 [95% CI, 73.0–100] % vs 88.9 [95% CI, 51.8–99.7] % for TB), and had 100% specificity for both diagnoses (100 [95% CI, 86.2–100] % and 100 [95% CI, 63.1–100] %), when considering data from all subgroups (Fig. 2A). Wide variations of p24 and CFP10 concentration were detected within the adult and child cohorts, but did not significantly differ between these groups (Fig. 2B and C). Serum p24 levels detected in samples of adults and children with p24-positive serum samples were all greater than the p24 LLMI, with median and interquartile ranges of 8.01 (6.44 to 13.47) and 11.96 (10.01 to 15.48) pmol/L. However, CFP10 levels detected in CFP10-postive serum samples in the adult and child cohorts primarily clustered near the CFP10 LLMI (median and interquartile ranges of 0.034 [0.022 to 0.042] and 0.040 [0.036 to 0.053] pmol/L, respectively). This included 2 samples from adults with TB that had CFP10 signals below the assay LLMI. None of the LC-MS/MS signals detected in serum of the HIV-1 negative or TB-negative individuals met the criteria for positive p24 or CFP10 signal. However, LC-MS/MS peaks detected in the serum of HIV-1-negative or TB-negative individuals had p24/Internal Standard (IS) or CFP10/IS peak area ratios that approached those detected in p24-positive and CFP10 positive samples. However, peak area overlaps were detected only in samples where the LC-MS/MS signal did not meet the criteria for positive p24 or CFP10 signal (online Supplemental Fig. 6).

Table 2.

Baseline characteristics of adult (TBAIT) and children (P1041) TB/HIV cohorts.

	TBAIT cohort (adult, n = 34)a,b				P1041 cohort (children, n = 17)a,b
	HIV+ (n = 14)	HIV− (n = 20)	TB+ (n = 23)	TB− (n = 11)	HIV+ (n = 9)	HIV− (n = 8)	TB+ (n = 9)	TB− (n = 8)
Female sex	5 (35.7)	7 (35.0)	5 (21.7)	7 (63.6)	6 (66.7)	3 (37.5)	6 (66.7)	3 (37.5)
Age (years for adult, days for child)	38 (27–42)	35 (27–42)	35 (27–42)	36 (31–41)	94 (92–98)	99 (92–109)	92 (92–98)	99 (94–109)
CD4 T count, cells/μL	195 (45–332) (n = 10)	—	225 (45–332) (n = 9)	165 (n = 1)	1530 (1257–1833)	—	1631 (1530–1833)	1305 (1120–1502)
CD8 T count, cells/μL	—	—	—	—	1227 (973–1350)	—	1227 (973–1325)	1166 (972–1417)
HIV RNA, Iog10, coples/mL	—	—	—	—	5.4 (4.6–5.9)	—	—	—

^aData was shown in n (%) or median (interquartile range, IQR)

^b“—”.

Fig. 2.

ILM p24/CFP10 assay results for clinical plasma/serum samples. (A), Plasma/serum p24pep and CFP10pep results in adults (n = 34) and children (n = 17) diagnosed with HIV-1/TB, TB only, HIV-1 only, and HIV-1-negative and TB-negative. Plasma/serum (B) p24pep and (C) CFP10pep concentrations detected in adults and children diagnosed as HIV-1/TB, TB only, HIV-1 only, or HIV-1-negative and TB-negative.

COMPARISON OF ILM AND IMMUNOASSAY FOR P24 DETECTION

Since multiple factors can affect the interval between HIV-1 infection and diagnosis by assays that detect serum levels of HIV RNA or HIV antigens and/or anti-HIV antibodies, we next compared the earliest positive results detected in serial post-HIV-1-exposure serum samples (n = 5) by the ILM p24 assay and several combined antigen/antibody assays commonly used for early diagnosis of HIV-1 infection. Serum ILM p24 assay results were positive a median of 5 days earlier than the best performing comparator antigen/antibody assay in 5 tested HIV-1 seroconversion panels, and a median of 5 days after the earliest HIV-1 RNA-positive result (Fig. 3A–J). Serum ILM p24 assay levels also revealed strong rank-sum correlation with the aggregate positive p24 immunoassay (r = 0.7576, Fig. 3K) and HIV-1 RNA (r = 0.9636, Fig. 3L) values of these samples. Notably, however, this analysis revealed a stronger correlation of ILM p24 results with HIV-1 RNA results than with p24 immunoassay results due to the results of one panel (#272), which yielded varying p24 trends when analyzed by different immunoassays (Fig. 3I). Rank-sum correlations between ILM p24 levels and positive p24 immunoassay and HIV-1 RNA values determined after exclusion of this data produced similar Spearman rank correlation coefficients (0.9643 and 0.9286; online Supplemental Fig. 7). Significantly, these excluded serum samples revealed a peak and decrease pattern when analyzed by the ILM p24 and HIV-1 RNA assays, but this was detected by only 1 of the 3 p24 quantitative immunoassays (Fig. 3I and J). These results suggested the potential utility of this ILM approach for real-time monitoring of HIV-1 viral burden.

Fig. 3.

ILM p24pep concentrations detected in longitudinal plasma samples spanning HIV-1 seroconversion interval of 5 HIV-1 patients (A-J), where the earliest positive HIV-1 RNA, p24 Ag/Ab immunoassay, and ILM positive results are indicated by blue, black, and red arrows, respectively. ILM correlation with immunoassay (K) and RNA (L) results.

ILM MONITORING OF P24 AND CFP10 RESPONSE TO TREATMENT

Multiplex measurement of p24 and CFP10 levels in plasma or serum by this ILM approach could represent a simple means to provide the integrated TB and HIV monitoring recommended by WHO guidelines. We therefore used ILM to measure CFP10 and p24 levels relative to antiretroviral therapy (ART) and/or anti-TB treatment (ATBT) initiation in serum collected from a small group of P1041 cohort children with confirmed HIV-infection who had available longitudinal serum samples and HIV-1 RNA concentration data (n = 5). Serum HIV RNA and p24 changes at the same or adjacent time points were compared in this analysis, but comparator data was not available to assess the agreement between serum CFP10 changes and systemic Mtb burden.

Serum p24 and CFP10 levels decreased following ART and ATBT initiation in 2 P1041-cohort children with HIV/TB diagnoses and positive ART and ATBT treatment responses, as indicated by serum HIV-1 RNA decreases and resolution of TB symptoms within 6 months of treatment initiation (Fig. 4A and B). Serum p24 and HIV-1 RNA levels detected at adjacent time points also revealed corresponding decreases in these cases. Conversely, 2 P1041 children who did not have TB symptom improvement or a decline in HIV-1 RNA after ART and ATBT initiation had corresponding CFP10 and HIV-1 RNA and p24 increases (Fig. 4C and D). Finally, one child demonstrated evidence of TB immune reconstitution inflammatory syndrome (TB-IRIS), defined as the paradoxical worsening of TB-associated symptoms/signs or radiological findings after effective ART initiation. Notably, this child had a CFP10 increase after ART initiation that matched an HIV decrease, although this RNA reduction was not fully matched by a p24 decrease. (Figure 4E).

Fig. 4.

Serum HIV-1 RNA (triangle) and ILM p24pep (circle) and CFP10pep (square) levels in HIV/TB-positive P1041 children with longitudinal serum who revealed (A and B) positive and (C and D) negative responses to ART and ATBT, or developed (E) TB-IRIS. Double-headed arrows and shaded area indicate ART and ATBT intervals and vertical lines indicate TB diagnosis (dashed) and TB cure (solid).

Discussion

The WHO recommends integrating services for patients with HIV and TB at the same visit to decrease TB-associated mortality (6). The ILM p24/CFP10 assay described in this study could aid this process, since it analyzes plasma/serum samples that can be readily obtained from all patient populations; uses a streamlined workflow suitable for high-throughput applications; employs immunoaffinity and MS to achieve high sensitivity and specificity; and quantifies 2 targets that should reflect the systemic burden of active HIV-1 and Mtb infection to enable rapid monitoring of ART and ATBT responses.

ILM p24/CFP10 assay results demonstrated good diagnostic sensitivity for HIV (p24) and TB (CFP10) in small adult and pediatric cohorts evaluated in this proof-of-concept study, revealing 85.7% (adult) and 88.9% (children) sensitivity and 100% specificity for HIV and 100% and 88.9% sensitivity and 100% specificity for TB. ILM p24 results did not attain the diagnostic sensitivity achieved by other HIV-1 tests in previous studies (98.8% to 100% sensitivity, 98.5% to 100% specificity) (21, 22), but may have been influenced by the limited size of the available testing cohorts and the use of cryopreserved biorepository samples. Serum ILM CFP10 results, however, exceeded the diagnostic sensitivity (≥66%) recommended for new non-sputum tests for microbiologically confirmed pediatric pulmonary TB (23), and diagnostic sensitivities reported with GeneXpert PCR assays in Mtb-culture-positive cohorts of young children (57%) or adults with HIV (79%) due to increased frequency of specimens containing low Mtb concentrations in these populations (24, 25).

ILM p24 results consistently detected early HIV-1 infection in longitudinal serum samples spanning HIV-1 seroconversion in 5 patients screened for HIV-1 infection. Positive ILM p24 results were detected within a median of 5 days before positive antigen/antibody assay and 5 days after positive HIV-1 RNA results, respectively, and these values correlated with the quantitative values of both assays. Serum p24 results can have a limited window for effective HIV diagnosis (15 to approximately 50 days post-infection) by standard immunoassays since p24 protein is rapidly bound and masked by host antibodies (17), although immune complex dissociation protocols can permit p24 detection beyond this point with reduced sensitivity. Serum ILM p24 assays should not be affected by this effect, however, since the trypsin digestion procedure used to generate the p24 target peptide should efficiently dissociate any endogenous p24 immunocomplexes.

ILM quantification of p24 and CFP10 in patient blood samples was also informative in assessing patient responses to ATBT and ART in this study, including TB-IRIS that can require rapid treatment modification to avoid life-threatening outcomes. Such cases can require different interventions if they represent a hyperinflammatory response to an existing Mtb infection (paradoxical TB-IRIS) or activation of a subclinical infection during ART-induced immune reconstitution (unmasking TB-IRIS) (26). However, TB-IRIS case definition remains challenging due to the lack of reliable diagnostic tests, and still primarily relies on a consensus clinical case definition (27). ILM analysis of serum p24 and CFP10 levels could provide direct evidence to identify TB-IRIS cases, which can be misdiagnosed as drug toxicity, drug resistance, poor adherence to treatment, or a result of opportunistic infection. Serum ILM CFP10 results could also be informative in evaluating the effectiveness of a treatment regimen to reduce Mtb burden, and real-time evaluation of this response could be used to detect Mtb clearance (signal loss) to provide evidence for a cure, Mtb persistence or growth (stable or increasing signal) that could indicate drug resistance, or signal TB recurrence (de novo signal after Mtb clearance) after treatment completion. These findings suggest this an integrated serum-based ILM p24/CFP10 assay approach has the potential to provide clinical information that could facilitate the coordination of TB and HIV diagnosis and treatment services to improve patient outcomes and merits further analysis in independent validation studies.

LIMITATIONS OF THE STUDY

Diagnostic performance estimates of the integrate ILM p24/CFP10 assay were generated using serum and plasma obtained from small proof-of-concept cohorts, and results from large well-characterized adult and pediatric HIV-1, TB, and HIV-1/TB patient and control cohorts are needed to provide more accurate diagnostic performance estimates. Similarly, estimates for the earliest diagnosis intervals of the 3 analyzed HIV-1 diagnostic assay types and correlations among their quantitative values also require larger cohorts to provide more accurate values, as does analysis of the HIV-1 and Mtb responses to ART and ATBT initiation. Finally, all these findings were obtained using a research LC-MS/MS system and further studies will be required to translate this assay to a LC-MS/MS system approved for use in clinical applications, if this assay is to be developed as a clinical assay rather than a laboratory developed test.

Nonstandard Abbreviations:

TB

tuberculosis

ILM

immuno-affinity liquid chromatography–tandem mass spectrometry

Mtb

Mycobacterium tuberculosis

CFP10

10-kDa culture filtrate protein

PLHIV

people living with HIV

mAb

monoclonal antibody

LLMI

lower limit of measuring interval

ART

antiretroviral therapy

ATBT

anti-TB treatment

IRIS

immune reconstitution inflammatory syndrome