: For example, if the normal range is 0-35 for a cancer antigen test, if a person’s test result is a number other than 0 (let’s say 20) , does that mean that they have cancer cells or that their risks of getting cancer are higher even though 20 is technically within “normal” range?
I’ve just always been curious as to how technicians are able to interpret these tests for accuracy to rule out cancer if the results don’t always say “0”.
Do most people have Cancer antigens?
In: 5
A test like this should come with interpretation guidelines either packaged with it or explained by a healthcare professional. How these guidelines are derived is typically published by manufacturers.
>I’ve just always been curious as to how technicians are able to interpret these tests for accuracy to rule out cancer if the results don’t always say “0”.
An example. I’m a scientist, and my lab develops testing procedures (“assays”). Back in 2020 we developed one to detect antibodies against a particular SARS-CoV-2 antigen in blood.
We first establish the *sensitivity* of the assay by testing increasingly diluted plasma of people who have had the virus (positive controls) as well as plain old water; this informs us about what values the assay reports when there is essentially nothing to truly measure — noise. Any result within about 2x those values at average (to be safe), we know is just nothing, and this gives us the absolute lower limit of what the assay can tell us. Let’s say this lower limit is about 0.1 unit.
Then we establish *specificity* by testing a lot of plasma from people who have definitely never had the virus (collected before 2019). We end up with a range of values, the vast majority of which are really low. For our purposes, 99% confidence is fine, so we take the 99th percentile value (i.e. the top 1%), and call that our cutoff for a positive result. Let’s say that value is 6 units.
Now, when we test unknown samples, we regard anything below 6 units as a negative result, and anything above that as increasingly positive (results commonly go into the multiple hundreds). Biological samples are always messy and have some potential for reactions that contribute to the measurement but are not the thing you are looking to measure. So in our assay, values of 1, 2, 5, 0.4… these are all *just negative*. The different values that the assay are reporting are within a range that we are very sure can be attributed to interfering reactions.
Hope this is informative.
Latest Answers