Elsevier

The Lancet

Volume 363, Issue 9418, 24 April 2004, Pages 1358-1363
The Lancet

Articles
A novel and accurate diagnostic test for human African trypanosomiasis

https://doi.org/10.1016/S0140-6736(04)16046-7Get rights and content

Summary

Introduction

Human African trypanosomiasis (sleeping sickness) affects up to half a million people every year in sub-Saharan Africa. Because current diagnostic tests for the disease have low accuracy, we sought to develop a novel test that can diagnose human African trypanosomiasis with high sensitivity and specificity.

Methods

We applied serum samples from 85 patients with African trypanosomiasis and 146 control patients who had other parasitic and non-parasitic infections to a weak cation exchange chip, and analysed with surface-enhanced laser desorption-ionisation time-of-flight mass spectrometry. Mass spectra were then assessed with three powerful data-mining tools: a tree classifier, a neural network, and a genetic algorithm.

Findings

Spectra (2–100 kDa) were grouped into training (n=122) and testing (n=109) sets. The training set enabled data-mining software to identify distinct serum proteomic signatures characteristic of human African trypanosomiasis among 206 protein clusters. Sensitivity and specificity, determined with the testing set, were 100% and 98·6%, respectively, when the majority opinion of the three algorithms was considered. This novel approach is much more accurate than any other diagnostic test.

Interpretation

Our report of the accurate diagnosis of an infection by use of proteomic signature analysis could form the basis for diagnostic tests for the disease, monitoring of response to treatment, and for improving the accuracy of patient recruitment in large-scale epidemiological studies.

Introduction

In the past 30 years, human African trypanosomiasis (sleeping sickness) has undergone a major resurgence in many areas of sub-Saharan Africa, and affects up to half a million people every year.1, 2 In northern Angola, epidemics of the disease cause a huge, but widely underestimated, morbidity and mortality of up to 50 000 cases every year.1, 2, 3 Despite its importance, human African trypanosomiasis is a neglected tropical infection, with few clinical studies on diagnosis and pathophysiology, and little attention paid to treatment programmes. In Angola, human African trypanosomiasis is a chronic, progressive disease caused by the parasite Trypanosoma brucei gambiense, and unless treated usually results in death. However, treatment of the disease is difficult, costly, and in 2–8% of cases lethal.1 Furthermore, the difficulty of diagnosing sleeping sickness is a major obstacle to effective management. Accurate methods of diagnosis would reduce rates of treatment-related mortality because hazardous treatments would be limited to those who really need them. Moreover, treatment costs would drop and efficiency of control measures in endemic areas would be improved.

Currently, human African trypanosomiasis can be accurately diagnosed only if symptomatic patients have detectable parasites in their blood, lymph, or cerebrospinal fluid (CSF).4 However, in West African trypanosomiasis, the parasite load (especially in the blood and CSF) is low and fluctuates in accordance with the antigenic variation of the parasite coat and host immune response.4, 5 Thus, parasitological confirmation often requires repeated examination of blood (>5 mL) and specialised concentration techniques, which are not always available in areas where human African trypanosomiasis is endemic. Therefore, when symptomatic patients from endemic foci present to sleeping sickness hospitals, diagnosis of human African trypanosomiasis relies on strongly positive serological evidence, such as the card agglutination test for trypanosomiasis (CATT), even if parasites cannot be visualised directly.6 This serological criterion for initiating treatment, however, remains controversial. False-positive CATT results (obtained without confirmation of titre) can occur in patients with schistosomiasis, filariasis, toxoplasmosis, malaria, or infections with non-pathogenic trypanosomes.7

In view of diagnostic difficulties, we aimed to investigate a proteomics-based approach to diagnose human African trypanosomiasis using serum samples.

Proteomic signature analysis was chosen because it has recently shown impressive accuracy when applied to the diagnosis of a variety of conditions, such as ovarian cancer,8 prostate cancer,9 breast cancer,10 and renal transplant rejection.11 This emerging technology is proving especially useful in diagnosing diseases where available tests are either too invasive or limited by poor diagnostic accuracy.

Section snippets

Participants

We gathered serum samples from 85 patients with human African trypanosomiasis presenting with symptoms to a dedicated sleeping sickness treatment centre—the Angotrip Treatment Centre in Ùige, Angola. Patients had one or both of: parasites visible in lymph, blood and/or CSF, or CATT titre ⩾1:8. These selection criteria were deemed sufficient to begin treatment for human African trypanosomiasis.

146 control serum samples were taken from various sources—patients and staff at Atkinson Morley's

Results

Participants' characteristics are shown in table 1. All patients with African trypanosomiasis were Angolan and CATT positive.

Spectra (2–100 kDa) from one sample run 28 times (seven assays) over 2 weeks gave coefficients of variation for peak size of 10·5% (intra-assay) and 20·6% (interassay). Coefficients of variation of mass to charge ratios, assessed with the same spectra, were ⩽0·05% (intra-assay and interassay). These coefficients of variation were calculated by averaging data from all

Discussion

Proteomic signature analysis accurately distinguished between serum samples from patients with human African trypanosomiasis and controls. Development of our technique has depended not only on proteomic technology, but also crucially on post-acquisition analysis of spectra. At present, the choice of spectral analysis algorithms by various investigators relies on personal preference, although such choices are known to have profound effects on the diagnostic accuracy.22 We adopted a novel

References (24)

  • QuY et al.

    Boosted decision tree analysis of surface-enhanced laser desorption/ionization mass spectral serum profiles discriminates prostate cancer from noncancer patients

    Clin Chem

    (2002)
  • LiJ et al.

    Proteomics and bioinformatics approaches for identification of serum biomarkers to detect breast cancer

    Clin Chem

    (2002)
  • Cited by (120)

    • Electronic health record for artificial intelligence health care, and application to liver disease

      2023, Artificial Intelligence, Machine Learning, and Deep Learning in Precision Medicine in Liver Diseases: Concept, Technology, Application and Perspectives
    • Prediction of fatty liver disease using machine learning algorithms

      2019, Computer Methods and Programs in Biomedicine
      Citation Excerpt :

      Artificial neural networks (ANN) are computational models that emulate the biological neural networks. It is very powerful nonlinear modelling which is already proven for accurate predictions in many CDS [13]. This model consists of a number of artificial neural units called “perceptron” [14].

    • Proteomics and Host-Pathogen Interactions: A Bright Future?. A Bright Future?

      2017, Genetics and Evolution of Infectious Diseases: Second Edition
    • Microbiological laboratory diagnostics of neglected zoonotic diseases (NZDs)

      2017, Acta Tropica
      Citation Excerpt :

      Furthermore, biomarkers strongly associated with ChD, have been identified in sera by the use of SELDI-TOF and matrix-assisted laser-desorption-ionisation time-of-flight mass spectrometry (MALDI-TOF MS) (Ndao et al., 2010; Santamaria et al., 2014). SELDI-TOF MS showed 100% sensitivity and 98.6% specificity for the use in patients with HAT (Papadopoulos et al., 2004). Comparable accuracy for identified ChD biomarkers via SELDI and MALDI-TOF has been reported (Santamaria et al., 2014).

    • Hematologic Changes Associated with Specific Infections in the Tropics

      2016, Hematology/Oncology Clinics of North America
    • Molecular diagnostics and parasitic disease

      2013, Clinics in Laboratory Medicine
      Citation Excerpt :

      For a more detailed discussion of HAT molecular assays, readers are referred to 2 recent reviews.115,131 Papadopoulos and colleagues31 recently described surface-enhanced laser desorption/ionization time-of-flight MS for diagnosis of HAT on the Ciphergen ProteinChip System Series PBS II (now Bio-Rad Laboratories, Hercules, CA), with reported sensitivities of 100%. This is a variation of matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) MS in which the sample bearing surface has intrinsic laser desorption ionization matrix properties and bioaffinity interaction capability.

    View all citing articles on Scopus
    View full text