Science1 publisher2 min readPublished
Belly button and nose odors differed by disease in a 24-person Fraunhofer pilot
Fraunhofer IIS researchers found swab odors from 24 people differed between healthy volunteers and those with Parkinson's, cognitive impairment or COVID-19. The team calls it a proof of concept, and a needle-free test would first have to hold up in larger, age-matched groups outside the lab.
The Scientist · Science desk
What happened
- The team swabbed the inner nose, outer ear and belly button, the body sites least exposed to perfume and soap.
- Samples went through laser-based photoacoustic spectroscopy, which identifies volatile compounds from the sound a sample gives off as it absorbs laser light.
- Of the three sampling sites, the nasal and belly button swabs gave the clearest results.
- Healthy participants and those with mild cognitive impairment had the most distinct odor profiles of the four groups.
- Only a few wavelengths across the measured spectrum carried the most important data.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- constraint Unmatched ages mean an odor difference between groups could reflect age as well as disease, and this cohort cannot separate the two.
- capability If the few-wavelength finding holds up in new people, an odor detector could be built around a handful of laser lines instead of the full spectroscopy setup.
- decision Teams developing needle-free odor tests get candidate swab sites and wavelengths to fix in advance for a larger study; the authors say this sample is too small for diagnostic conclusions.
Each group in the study had six people: six with Parkinson's disease, six with mild cognitive impairment, six with COVID-19 and six healthy controls [4]. Every condition is therefore a quarter of the cohort, and 18 of the 24 participants had one of the three conditions [1]. According to Discover's account of the Scientific Reports paper [1], the molecular profiles showed patterns that matched among people with the same condition [6]. The article does not give a classification accuracy, or say whether the patterns were checked on people left out of the original analysis.
The choice of sampling sites is the part of the design I like. Swabbing the places that see the least perfume and soap removes one contaminant at the source [3]. Others remain. Volatile organic compounds, the molecules being measured, vary with diet and lifestyle as well as with illness [2], and a group of six leaves little room to average that variation out.
The authors list the lack of age-matched groups among their limitations [11]. Without matching, a profile that separates patients from controls could be picking up age alongside disease, and 24 people cannot pull the two apart. That caveat applies to the study's sharpest contrast, between the healthy and cognitive impairment groups [8], as much as to any other.
The few-wavelength result [9] matters most to anyone thinking about hardware, and I would treat it most cautiously. Choosing a handful of informative wavelengths from a wide spectrum, with 24 people, is the step in an analysis most likely to fit noise. Those wavelengths would need to be fixed in advance and tested on a new cohort before they say what a clinic instrument should measure. The measurements were also made under highly controlled laboratory conditions that, the team notes, are not representative of real-world testing [12].
The authors' own summary is measured. They wrote that "overall, the findings of this study demonstrate the feasibility of LPAS-based headspace analysis for detecting disease-associated spectral patterns" [13].
What to watch
- A replication in a larger, age-matched cohort that tests the same swab sites and pre-specified wavelengths on new participants.
- Whether the team reports classification accuracy on participants held out of the analysis that found the patterns.
- A reduced instrument built around the few informative wavelengths, tried outside controlled laboratory conditions.