3. Results
The demographic characteristics of the participants in the study,
includes 29 patients diagnosed with CCHF and 10 healthy individuals
serving as controls. The mean age of the control group was 50.1 years,
with an age range between 40 to 64 years, while the CCHF patient group
had a slightly higher mean age of 50.5 years, though with a wider age
range of 22 to 77 years. In total 126 blood samples were collected,
their serum extracted by methanol extraction, and NMR samples were
prepared for all samples. Following metabolite extraction, NMR analyses
were conducted on the samples. Finally, a statistical analysis approach
was applied. In the statistical analyses, all samples collected over a
period of 4 days. All groups were subjected to statistical evaluation as
moderate, severe, and control groups as described in materials and
methods.
In PLS-DA analysis, a model is developed to include the primary
components (latent variables) of the data. The first five components are
considered important in classifying the data and explain the majority of
the dataset. The matching score plots exhibit these first five
components in two-dimensional graphs and compare each component with
another (Figure 1a). Subsequently, the next step involved obtaining the
Partial Least Squares Discriminant Analysis (PLS-DA) score plot, as
shown in Figure 1b demonstrates a high discrimination between healthy
and diseased individuals. PLS-DA effectively deals with highly
dimensional data, highlighting the relevant variables responsible for
the distinction, thereby providing a clear differentiation between
health states based on the variation and correlation within the
biomarker data. In order to examine the significant data packets
responsible for this discrimination, the PLS-DA VIP Projection variable
importance score plot is illustrated in Figure 1c. This technique,
graphically demonstrated in Figure 1c, reveals the variables that have
substantial influence on the discrimination. VIP scores signify the
importance of each variable in the model, enabling us to identify and
focus on the key biomarkers that are significantly altering between
healthy and diseased states. This plot aids in the in-depth
interpretation of the complex multivariate data and fosters a more
nuanced understanding of the disease markers. Figure 1c showcases the
top 15 variables that have the greatest impact on discrimination among
these variables. As shown in Figure 1c, it is noticeable that several
data packets exhibit low concentration values (blue) in the control
group and gradually increase towards higher concentrations with the
passage of time. Ultimately as depicted in Figure 1d, the listed
compounds exhibited a notable increase in their concentration, as
substantiated by increased signals from their corresponding data
buckets. Using a statistical analysis approach consisting of three
different groups. Samples taken on four consecutive days from the two
patient groups were successfully differentiated from the control group.
However, no discernible distinction was observed between the moderate
and severe groups. Overall these compounds were detected to be elevated
compared to the control: SAH, GTP, Carnosine, maleate, 2-Deoxyuridine,
IMP, AMP and NADP+.
As a secondary approach to data analysis, data collected on the day 1
and day 2 were used as severe patient and moderate patient groups to
further investigate the metabolite profile of the patients during the
initial stages of hospitalization. Subsequently, only samples obtained
on day 1 and day 2 were categorized as diseased samples, and the blood
samples obtained from healthy individuals were designated as the control
group for statistical analysis (Fig 2). In addition to compounds
detected in our initial analysis novel compounds with significant
increase during day 1 and 2 detected that may have a more active role in
the patogenesis.
Finally, in a comprehensive metabonomic analysis of CCHF patients’ blood
serum, distinct patterns of metabolites were observed between the severe
and moderate infection levels. Specifically, through the PLS-DA score
plots and Variable Importance Score (VIP) projections detailed in Figure
3, notable variations in the metabolite profiles were discerned. In the
case of severe CCHF infections, the compounds AMP, IMP, and NAAD were
identified as present within the serum, whereas they were not as
significantly detected in the samples from patients with moderate
infection levels. Conversely, GTP was found to be significantly
increased in the samples of patients with moderate CCHF infection
levels, but not in those categorized as severe. Overall the results
outlined in our final analysis reveal a marked metabolic variety between
the moderate and severe CCHF infection levels which may indicate the
critical pathways in CCHF patogenesis.