Abstract:
Background: Metabolic complications resulting from the human immunodeficiency virus
type 1 (HIV-1) and the acquired immunodeficiency syndrome (AIDS) are as common as the
immune system disruption caused by the virus, but not as well known. Highly active
antiretroviral therapy (HAART) used to treat HIV-1 infection exacerbates the effects HIV-1
has on the host‘s metabolism. Common metabolic complications such as insulin resistance,
lipodystrophy, lactic acidosis and others contribute to morbidity and mortality during
HIV/AIDS. The detection of HIV-1 related metabolic biomarkers assists in diagnosing and
monitoring metabolic complications, however, limitations of the conventional methodologies
used for detecting these molecules caused a paucity of data on HIV-related metabolic
indicators. Metabonomics, the ability to measure multiple metabolites simultaneasously,
shows promise in distinguishing HIV-1 negative and positive patients through nuclear
magnetic resonance (NMR) and vibrational spectroscopy as well as mass spectrometry (MS)
profiles of various biofluids. The objective of this study was to determine the abilities of
ultra-performance liquid chromatography-mass spectrometry (UPLC-MS) in the
identification of serum metabolites associated with HIV and/or HAART. The study was also
aimed at determining whether data from a less sensitive technique, Fourier transform infrared
(FTIR) spectroscopy would lead to comparable conclusions as those derived using UPLC-MS
data.
Methods: Sera were collected from three experimental groups; HIV negative (n=32), HIV
positive (n=29) and HIV positive patients receiving HAART (n=34). Metabolites were
extracted using a conventional approach of cold methanol extraction as well as the OstroTM
plate extraction technology which involved filtration by positive pressure. The filtrate was
analysed in the negative and positive electrospray ionization (ESI) modes of UPLC-MS.
Serum samples were also dried overnight and analysed using FTIR. Data processing and
chemometric analysis was carried out using the SPSS 19.0 and MassLynx v4.1 software packages. Following extensive statistical evaluation of data, bioinformatics approaches that
assisted with metabolite identification were conducted.
Results: The combination of OstroTM plates and UPLC-MS produced high resolution
chromatograms that showed visible differences among the serum samples of HIV negative,
HIV positive and HIV positive patients receiving HAART. Linear discriminant analysis
(LDA) classified experimental groups into the correct categories with great accuracy (>88%),
using potential biomarkers responsible for the observed group variations. Principal
component analysis (PCA) showed clear separations as well as some overlap among the three
experimental groups. Orthogonal projections to latent structures-discriminant analysis
(OPLS-DA) showed clear differences between two classes of samples at a time and potential
biomarkers were selected from accompanying S-Plots. Hundred and twelve distinct group
distinguishing metabolites detected from both ESI positive and ESI negative modes were
significantly altered (p<0.05). HIV and/or HAART altered metabolites of energy, neuronal
and mitochondrial processes were identified and were evident in the amino acid,
carbohydrate, lipid and nucleoside/nucleotide metabolic products being detected. Antiviral
drugs [mostly nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs)], antiinflammatory
and anti-cancer drugs, nutrient supplements and other drugs associated with the
treatment of HIV conditions also contributed to class variations. FTIR generated metabolic
patterns that separated the three experimental groups on an LDA scatter plot which achieved
correct (>87%) classification accuracy. The significantly altered patterns indicated hydroxyl
and alkene group vibrations and these groups were present in the metabolites identified by
MS.
Conclusion: OstroTM plates and UPLC-MS successfully purified, detected and identified sera
metabolites distinguishing HIV and/or HAART patients. The different statistical analysis
methods applied in this study were in agreement and the OPLS-DA statistical tool
complemented the sensitivity of UPLC-MS for the detected distinguishing metabolites. The
approach employed here delivered promising findings for use in the discovery of metabolic
biomarkers. Distinguishing metabolites identified could be traced to HIV-infection and/or
treatment. Findings from this study corroborated with others which showed that NRTIs
remain a challenge in the era of HAART toxicities, especially their dominant effect on
mitochondrial dysfunction. This work therefore suggests the use of UPLC-MS in HIV disease
diagnosis, prognosis, monitoring of treatment success or failure and the ability to link treatment to metabolic complications. Even though FTIR is less sensitive than UPLC-MS, it
was successful in detecting metabolic patterns that corresponded to some metabolites
detected by UPLC-MS. This suggests that this easier to perform technique also has potential
clinical application in monitoring HIV/AIDS.