Having initially trained as a medical technologist performing assays in clinical hospital labs using similar immunoassay techniques like is done for determining the value of CA19-9, one had to be versed in the limitations of the assay, interfering substances, intra and inter variability of a particular test method. There are a number of manufacturers of the CA19-9 assay (Roche, Abbott, Siemans) and many reference laboratories and large medical centers where there is a high volume of patients to justify doing the test in-house. Each manufacturer has determined by statistical analysis what the variability and error of measurement is for their test kit.

A variation of +/- 2 is insignificant. During my clinical research careeer, I worked on developing a number of assays of blood and urine analytes and remember all to well having to do a substantial amount of statistical analysis to determine variation in results called standard error of the mean. Samples were run in triplicate….something that is not likely done in clinical testing vs when a test is being developed and optimized.

During the pandemic I had to have my CA19-9 done at a large nation testing lab instead of the medical center I go to for exams and surveillance. The value was 5 points higher. When I went back to the hospital lab for testing, my value was what it always measured at. Ever ask a nurse to recheck your blood pressure measurement when the first result seemed higher or lower than what you normally run? I have and usually the second measurement falls more in line with what I normally run. That is the problem when a test value of n=1 is used.

A rise of 2U/nL of CA19-9 is within the margin of error of the CA19-9 assay. It would be interesting to see what the values would be had your sample been run in triplicate and whether two of the three matched with one outlier or they all matched. The change in value is also not statistically significant to indicate a change in disease treatment status. There are also benign conditions that can cause variations…from inflammation to the use of the vitamin B-7 commonly known as Biotin contained in both pharmaceutical and cosmetic products. The next test point is the one that will give an answer and there needs to be a significant increase to indicate concern is warranted.

This link to Quest Diagnostics indicates a margin of error of 3U/nL.
https://testdirectory.questdiagnostics.com/test/test-guides/TS_CA_19-9/ca-19-9-serum?p=td