23

Clearly the test line binds some component of the coronavirus, I believe most commonly the nucleocapsid protein (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8049777/). However, despite extensive googling (and Google is doing a lousy job of understanding my question) I have not been able to figure out what is being bound by the control line. Probably it differs by test manufacturer, but I'm sure there are some relatively common antibodies that are used in the control line?

My best guess is that it should test for some common mucin or collagen or elastin but this is based on nothing.

The reason I'm asking is that I want to understand if the "control" line being present is a good indicator of the test having been done properly. Let's say some person:

  • rubs the swab on a different body part than the nasal mucosa
  • or doesn't touch the body at all with it
  • or swabs the nasal mucosa very briefly
  • or only inserts the swab into the buffer solution extremely briefly

would the control line still show up?

I'm not simply trying this myself at home because it's annoying and expensive to get lots of tests simply to waste them to test their limits.

Hinton
  • 333
  • 2
  • 5
  • Related: https://biology.stackexchange.com/questions/101767/whats-happening-in-the-c-and-t-stripes-of-a-covid-test-kit/101777#101777 – canadianer Sep 29 '22 at 17:25

1 Answers1

25

According to biotrend (SARS-CoV-2 (Covid-19): Rapid Antigen Test for Diagnosis), the control line has an immobilised Goat anti-chicken antibody. There, a chicken-IgY gets stuck which was carried from a conjugate pad by the sample fluid.

So even pure buffer solution would yield a positive signal at the control line. It seems that the control only tests for proper reactivity of the antibodies contained in the conjugate pad, and not for proper usage by the user:

A colored band in the control region serves as a procedural control, indicating that the appropriate sample volume has been added and the membrane is functioning.

This manual (health.govt.nz) reports that whole blood and Mucin do not interfere with the test (including control?). It sounds as if biological fluids are meant to be ignored by the test in general.

markur
  • 1,769
  • 4
  • 21
  • 2
    Indeed, this is why the control line is so stupidly designed. It should check not only for proper functioning of the test but also that the user did not just put pure buffer solution! Otherwise a negative test is not convincing even if supervised. – user21820 Sep 30 '22 at 05:02
  • 4
    @user21820 No, this is not stupid because you need a control that shows that your test is valid which is independent of the sample. Having a positive control would sometimes be nice but makes the interpretation more complicated. Not to speak of the need to have a control that under no circumstances will produce a false positive result. – Chris Sep 30 '22 at 05:50
  • 4
    @Chris: You claim "you need". I disagree. What we need from the test is a way to distinguish between covid infected and non-infected. The way the tests have been designed is inferior for that purpose. For what I suggested, there is no complicated interpretation at all. (1) No control line ⇒ Test invalid. (2) Control line ∧ Test line ⇒ Infected. (3) Control line ∧ No test line ⇒ Likely uninfected. – user21820 Sep 30 '22 at 05:56
  • 1
    @user21820: yes, in theory that sounds good. However, using a user dependent positive control introduces another variable, where the test may fail (making the test look worse compared to others). Also, it wouldn‘t ensure proper usage, since it does not prevent a user from applying an older/someone else‘s healthy sample. Supervision or Conduction by professionals is the only solution. I think it‘s smart. – markur Sep 30 '22 at 06:08
  • 6
    @user21820 Believe me, you need this. What do you do with tests where the positive control is absent but the test band is present? Also: How do you control if the test is working at all? Antibodies are stable, but still storing conditions are important to ensure a result. – Chris Sep 30 '22 at 06:20
  • 4
    @markur Test line obviously means you have COVID. Control line missing obviously indicates you did something wrong with the sample. Intuitive conclusion: you did something wrong, but COVID was detected anyway; if that bothers you, take another test. – user253751 Sep 30 '22 at 22:59
  • 1
    @user253751 - I know, I‘ve been there, like everyone else. My point was that a user dependent positive control doesn‘t prevent fraud, hence it‘s worth can be debated. (For me I‘d feel more comfortable with sth like a Mucin control, but it‘s really not that necessary, given that these tests are expensive enough) – markur Oct 01 '22 at 07:16
  • 1
    @Chris: What user253751 said... No control line ∧ Test line ⇒ Probably infected but you should get another test done because you botched the first one. If you keep botching tests unlike everyone else, then go to a hospital or clinic. – user21820 Oct 01 '22 at 08:51
  • 3
    @markur: This isn't just about malicious testing. Even an honest person can't be sure that they have swabbed enough if the control line doesn't give them any feedback. – user21820 Oct 01 '22 at 08:54
  • 3
    @user253751: Test line appearing but no control line does not mean you have COVID. It means something went wrong that caused colloid to deposit at the test stripe, but not at the control stripe. You can say in that situation that the test is unreliable, but not more. The control is meant to guard against false negatives due to errors in the manufacturing process or due to storage conditions. – cbeleites unhappy with SX Oct 01 '22 at 13:36
  • @user21820: sure "non-contact swabbing" will cause false negatives. But: even proper swabbing (taking up sufficient mucin) is subject to sampling error due to spatial heterogeneity. And there is indication that this is a substantial source of error (I've seen some very early paper in 2020 looking at a series of PCR results from back-of-the-throat swabs of the same individual, and roughly a quarter of those was negative. The rapid test sampling procedures like nasal swab are considered to have sufficient sensitivity in practice compared to that standard. The question is: compared to that, ... – cbeleites unhappy with SX Oct 01 '22 at 13:44
  • ... would an additional mucin control line be worth while? In particular, since a doubt about sufficient swabbing (e.g. by someone overseeing the test) can be resolved by doing another test (from the same lot). But a manufacturing or storage problem is likely to affect all kits of the lot or the boxes at hand. – cbeleites unhappy with SX Oct 01 '22 at 13:46
  • For the FlowFlex brand, apparently substitute "mouse" for "chicken". Are these likely to be in-vitro manufactured antibodies/antigens, or is someone extracting them from actual goats/chickens/mice? – Daniel Hatton Oct 01 '22 at 16:57
  • 1
    @cbeleitesunhappywithSX: I didn't say you must have covid. That's why I used the word "probably". The likelihood of a manufacturing error is much lower (<0.1%) than the likelihood that the test line shows up positive but the person does not have covid. Yes, an entire batch may be faulty, and that's why I said "should get another test done". – user21820 Oct 01 '22 at 17:47
  • @user21820: no, that was user253751, to whom also my comment went. Still, you're not doing the comparison we need here. In particular, we'd need the (posterior) probability of the SAR S-CoV2 line of the test being correctly positive given that the control line did not show up. Your < 0.1 % for manufacturing error is only the the prior (BTW do you happen to have a reference on this - this is not to raise doubt on the number in case it's a guesstimate, it's that I'm professionally interested in any such ref), but the data (control not showing up) is actually very strong evidence here... – cbeleites unhappy with SX Oct 02 '22 at 09:55
  • (prior of manufacturing/storage error: I didn't see that many test kits, all in all ≈100, but there were some in there which were obviously short on extraction buffer). For the practice of covid testing, the 2 € for a test kit makes the discussion moot, since the practical answer will always be throw it away and do another one. But we're on biology.sx here where other lab tests are discussed as well, and I'm analytical chemist: it is an important point to me that an invalid test is invalid and it is wrong to use it for anything else than possibly digging into fault modes. – cbeleites unhappy with SX Oct 02 '22 at 10:05
  • 1
    @DanielHatton these are most likely to be recombinant anti-species antibodies, as their performance is much more similar lot-to-lot than animal-derived polyclonal antibodies. – MattDMo Oct 02 '22 at 23:11
  • @cbeleitesunhappywithSX: There's clearly a difference in our goals. In my view, the purpose of using the ART is not to get a near-zero false-positive rate, nor to get reliable statistics of covid infection. Rather, it is to provide confidence that you do not have covid based on the ART result! If you have that goal in mind, then you will understand why I say that my suggested version is better, and why we do not need the conditional probability of infection given no control line but a positive test line. – user21820 Oct 03 '22 at 08:43
  • If you follow my stated procedure (including getting another test done if the first one has a problem), then that probability is irrelevant. Rather, it is the probability of manufacturing error that is relevant, because that would give an upper bound on how likely there is a problem with the ART in the first place. You can't just treat this ART testing as a lab test. It is not, because its primary purpose is as a tool for dealing with the pandemic. Whenever I use the ART, I want to know ( likelihood of covid given negative ART result ), not ( likelihood of no covid given positive ART result )! – user21820 Oct 03 '22 at 08:52
  • Thanks @MattDMo. Same for the mouse/chicken antigen? – Daniel Hatton Oct 03 '22 at 11:17
  • 2
    @DanielHatton yes, they're all recombinant. However, recombinant doesn't necessarily mean monoclonal - the final product could be a mix of two or more different clones to give the desired characteristics to the batch. I used to work at an antibody company that did this. – MattDMo Oct 03 '22 at 11:35