# Biospectroscopy Combined with Multivariate Analysis as Tools for Identifying Trypanosoma cruzi Discrete Typing Units in Triatoma brasiliensis (Hemiptera: Reduviidae: Triatominae)

## TL;DR

This study shows that infrared spectroscopy and data analysis can quickly and noninvasively detect and classify Trypanosoma cruzi infections in Triatoma brasiliensis insects.

## Contribution

The integration of ATR-FTIR spectroscopy with chemometric models offers a novel noninvasive method for T. cruzi detection and genotyping in triatomine bugs.

## Key findings

- Classification models achieved 100% sensitivity and specificity for TcII and mixed infections.
- Spectral differences in proteins and nucleic acids enabled discrimination between infected and uninfected insects.
- PCR confirmed 92.5% of infections detected by the spectroscopy method.

## Abstract

Epidemiological surveillance of Chagas disease with determination
of Trypanosoma cruzi (T. cruzi) positivity and genotyping is important
for the adoption of control measures. The association of biospectroscopy
with chemometric models can be used as an alternative tool to determine T. cruzi positivity and genotyping directly on the
insect, with laser incidence, without damage to the insects, and without
parasite isolation and growing. In this study, infrared spectroscopy
(ATR-FTIR) combined with classification and authentication models
was used for the identification of the experimental infection of Triatoma brasiliensis (T. brasiliensis) by different discrete typing units of T. cruzi. T. brasiliensis fourth and fifth
instars were experimentally infected with 40,000 parasites (TcI, TcII,
TcIII, or mixed infection) per milliliter of blood, and 1, 15, and
30 days after infection, infrared spectra were collected from the
abdomen of each insect. The classification models, genetic algorithm
linear discriminant analysis (GA-LDA) and successive projections algorithm
linear discriminant analysis, achieved 100% sensitivity and specificity
for TcII, the mixed infection, and control across all infection periods.
For TcI and TCII, GA-LDA performed best in 15 days with 75% sensitivity
and 94% specificity. The data-driven soft independent modeling of
the class analogy model correctly classified most infected samples
within the limits of both the training set and the test set, excluding
the uninfected samples of the acceptance region, and achieving sensitivity
and specificity close to 100%. Spectral differences, primarily attributed
to proteins (amide III band, 1247–1307 cm–1) and nucleic acids (phosphate stretching vibrations, 1048–1085
cm–1), allowed for consistent discrimination between
infected and uninfected insects. The polymerase chain reaction of
kDNA analysis confirmed 92.5% (37/40) of infections in all triatomines
submitted to experimental infection. Thus, the integration of ATR-FTIR,
classification tools, and authentication models has been shown to
be a rapid, noninvasive, and promising approach for the diagnosis
and entomological surveillance of Chagas disease.