IntelligentMaterialSolutions
Use case · Diagnostics research

Tuberculosis on a lateral-flow strip.

Researchers at Leiden University Medical Center and the Spallanzani institute in Rome measured seven proteins from the body’s own immune response, one paper strip per protein. The test told active TB from latent TB and from COVID-19, and tracked patients through treatment. The light on every strip came from IMS crystals.

iScience · January 20237 host proteins · 7 stripsOpen access · CC BYResearch use only
Read the paper ↗How the strip works
The problem

TB tests still start with sputum.

Active TB is diagnosed by finding the bacterium in sputum, the mucus coughed up from the lungs. The paper sets out why that is a bottleneck.

Hard to collect

Sputum is difficult to obtain, especially from children, and has a large sampling error that leads to false negatives.

Slow and lab-bound

The methods are often expensive and time-consuming and need trained staff. Extra clinic visits before treatment risk losing patients to follow-up.

COVID-19 looks similar

The two diseases can share signs and symptoms, and the pandemic disrupted TB services. A fast way to tell them apart helps triage.

WHO figures cited in the paper: in 2020 around 10 million people developed TB and 1.5 million deaths were attributed to it.

The idea · how the strip works

Measure the body’s response, not the germ.

Instead of looking for the bacterium, the team measured seven proteins the immune system makes in diluted blood serum. Each protein has its own 4 mm strip. A benchtop reader scans each strip and reports one number: the ratio of light at the Test line to light at the Flow-Control line.

IllustrationNot to scale · readout values illustrative
Animated lateral-flow strip: diluted serum flows along the strip, IMS crystals coated with antibodies pick up the target protein and stop at the Test line, leftover crystals stop at the Flow-Control line, then an infrared reader beam makes both lines glow and the reader reports the Test to Flow-Control ratio.
Diluted serum is taken up by the strip.
IMS crystal carrying antibodiesTarget protein from the serumCrystal glowing under infrared
  1. 1
    Dilute and dipSerum is diluted in buffer, from 10-fold to 10,000-fold depending on the protein, and the strip is placed in the well.
  2. 2
    IMS crystals carry the antibodiesAntibodies are bound to IMS upconverting crystals in the pad. As the liquid moves up the strip, the crystals pick up the target protein.
  3. 3
    Two lines catch the crystalsThe Test line holds a second antibody that catches crystals carrying the protein. The Flow-Control line catches the crystals that pass by.
  4. 4
    Infrared in, visible light outOnce the strip is dry, the reader shines infrared light. Only the crystals answer, with visible light. Result = Test signal ÷ Flow-Control signal.
The reporter on every strip

“Polyacrylic acid functionalized UCPs (200 nm, NaYF4:Yb3+,Er3+; Intelligent Material Solutions Inc., Princeton, NJ, USA)” Methods, Pierneef et al. 2023

ApoA1Apolipoprotein A1
CRPC-reactive protein
FerritinIron-storage protein
IL-6Interleukin-6
IP-10Chemokine CXCL10
SAA1/A2Serum amyloid A
S100A12Calcium-binding protein
Key results, as reported

Seven strips, three questions.

For each question the team counted how many markers were above a cut-off and called the result positive above a set count. Sensitivity is the share of patients the test caught; specificity is the share of the other group it correctly cleared; AUC runs from 0.5 (a coin toss) to 1.0 (perfect).

Active TB vs latent TB
83%sensitivity97%specificity

AUC 0.94 · 6-marker score, 3 or more positive. CRP + SAA1/A2 alone: AUC 0.91, same 83% / 97%.

TB n = 30 · latent TB n = 29
Healthy vs COVID-19
93%sensitivity100%specificity

AUC 0.99 · 7-marker score, 4 or more positive.

Healthy n = 39 · COVID-19 n = 102
TB vs COVID-19
91%sensitivity87%specificity

AUC 0.95 · 7-marker score, 4 or more positive. CRP + SAA1/A2 + S100A12, 2 or more: 90% / 87%, AUC 0.94.

TB n = 46 · COVID-19 n = 102
Treatment monitoring
TB · n = 22Months 5–9 of treatment: IL-6, ferritin, CRP and S100A12 fell; ApoA1 rose.
COVID-19 · n = 25About 6 weeks after discharge: CRP, ferritin, IL-6, IP-10, SAA1/A2 and S100A12 fell; ApoA1 rose.

All results are in-sample. The authors write that the markers still need validation in an independent cohort recruited at the same site and time.

Strip by strip

See which lines brighten and dim.

Pick a comparison. On each strip pair, the left strip is the first group and the right strip the second. The right Test line brightens when the paper reports a higher level, dims when lower, and matches when there was no significant difference.

Latent TB (n = 29) vs active TB (n = 30)

Six of the seven proteins were significantly higher in active TB. ApoA1 showed no significant difference. A score counting six markers (ApoA1, CRP, ferritin, IL-6, IP-10, SAA1/A2), positive at 3 or more, gave sensitivity 83%, specificity 97%, AUC 0.94. Two markers alone, CRP + SAA1/A2, gave AUC 0.91 with the same 83% / 97%.

S100A12 was developed later and tested on 29 latent and 26 TB samples.
ApoA1Apolipoprotein A1
No significant difference
AUC 0.57
CRPC-reactive protein
Higher in TB
AUC 0.87
FerritinIron-storage protein
Higher in TB
AUC 0.66
IL-6Interleukin-6
Higher in TB
AUC 0.76
IP-10Chemokine CXCL10
Higher in TB
AUC 0.70
SAA1/A2Serum amyloid A
Higher in TB
AUC 0.87
S100A12Calcium-binding protein
Higher in TB
AUC 0.96

Illustration Directions as reported; line brightness is illustrative. AUC values from Table 2 of the paper; p values from its Results.

Show all comparisons as a table
ComparisonApoA1CRPFerritinIL-6IP-10SAA1/A2S100A12
Latent TB vs TB— n.s.
AUC 0.57		▲ higher
AUC 0.87		▲ higher
AUC 0.66		▲ higher
AUC 0.76		▲ higher
AUC 0.70		▲ higher
AUC 0.87		▲ higher
AUC 0.96
Healthy vs COVID-19▼ lower
AUC 0.85		▲ higher
AUC 0.94		▲ higher
AUC 0.96		▲ higher
AUC 0.72		▲ higher
AUC 0.82		▲ higher
AUC 0.98		▲ higher
AUC 0.89
TB vs COVID-19▼ lower
AUC 0.71		▲ higher
AUC 0.93		▲ higher
AUC 0.65		— n.s.
AUC 0.58		— n.s.
AUC 0.55		▲ higher
AUC 0.92		▲ higher
AUC 0.79
Treatment: TB▲ higher
p = 0.0133		▼ lower
p = 0.0133		▼ lower
p = 0.0002		▼ lower
p = 0.0208		— n.s.
p = 0.1409		— n.s.
p > 0.9999		▼ lower
p = 0.0004
Treatment: COVID-19▲ higher
p = 0.0063		▼ lower
p < 0.0001		▼ lower
p = 0.0001		▼ lower
p < 0.0001		▼ lower
p = 0.0005		▼ lower
p < 0.0001		▼ lower
p = 0.0022

“Higher” and “lower” describe the second group against the first (for treatment: later against before). n.s. = no significant difference.

Why our crystals

A reporter built for numbers.

The strip reads out a ratio, not a yes or no. That only works if the reporter is clean, steady and the same every time.

How the crystals workIP & patents
  • No background glowInfrared goes in, visible light comes out. Serum, paper and plastic do not do that, so only the crystals light up.
  • They do not fadeInorganic crystals do not bleach under the reader’s light, so the signal holds while the strip is scanned.
  • QuantitativeThe light from a line tracks the crystals on it, so the reader gives a number: the Test to Flow-Control ratio.
  • Uniform sizeCrystals grown to one size behave the same in every strip, so one batch gives the same response from strip to strip.
Patent · co-owned with LUMC

U.S. 10,054,593 · Multiplexed spectral lifetime detection of phosphors. Co-owned by Intelligent Material Solutions, Leiden University Medical Center and the U.S. Department of Health and Human Services. The inventors include Paul Corstjens. See the IP & patents page.

Where this stands

Research use. Not a cleared test.

  • Research study. The strips were used on banked and study serum samples in Europe. They are not a cleared diagnostic.
  • In-sample results. The authors call for validation in an independent cohort, with TB and other respiratory diseases recruited at the same site and time.
  • Severe COVID-19 only. The COVID-19 patients were hospitalized in 2020 and severely ill, admitted at different stages of disease.
  • An aid, not a verdict. These host proteins are not specific to one disease. The authors see the strips as a triage and adjunct tool, read alongside the patient’s symptoms and local disease burden.
The paper · open access iScience 26, 105873 · January 20, 2023 · CC BY 4.0 Host biomarker-based quantitative rapid tests for detection and treatment monitoring of tuberculosis and COVID-19 Pierneef L, van Hooij A, de Jong D, Tjon Kon Fat EM, van Meijgaarden KE, Petruccioli E, Vanini V, Roukens AHE, Goletti D, Corstjens PLAM, Joosten SA, Geluk A, with the BEAT-COVID study group. Leiden University Medical Center, the Netherlands, and National Institute for Infectious Diseases “L. Spallanzani” IRCCS, Rome, Italy.
doi.org/10.1016/j.isci.2022.105873 ↗All papers by Paul Corstjens and colleagues →Lateral flow with Intelligent Material →

Research use only. Not a cleared diagnostic.

Building a quantitative rapid test?

We supply upconverting reporter crystals to research teams and test developers. Tell us what you want to measure.

Talk to us about diagnosticsSelected papers