Accepted answer
Three tests, three questions, no overlap. And your suspicion about "sterile filtered" is correct: it is a process description, not a test result, and it appears on a COA in the space where a result should be.
| Bioburden | Sterility | Bacterial endotoxin |
|---|
| Question answered | How many viable organisms are present? | Are there zero viable organisms? | How much Gram-negative cell-wall lipopolysaccharide is present? |
| Result type | A count (CFU/g or CFU/mL) | Pass or fail | A number (EU/mg or EU/mL) |
| Compendial chapter | USP 61 and 62, EP 2.6.12/2.6.13 | USP 71, EP 2.6.1 | USP 85 and 86, EP 2.6.14 and 2.6.32 |
| Turnaround | 3 to 5 days | 14 days minimum | Under 2 hours |
| Destructive? | Yes | Yes, and it consumes many containers | Yes, but micrograms |
| Detects dead organisms? | No | No | Yes — this is the point |
| Detects fungi/yeast? | Yes, as TYMC | Yes, via SCDM at 20 to 25 degrees | No |
| Detects Gram-positive contamination? | Yes | Yes | No — Staphylococcus produces no endotoxin |
| Survives autoclaving? | n/a | n/a | Yes, almost entirely |
Bioburden
A plate count. Dissolve or suspend a weighed sample, plate on soybean-casein digest agar for total aerobic microbial count and on Sabouraud dextrose agar for total combined yeasts and moulds, incubate, count colonies. It is a quantitative test applied to material that is not claimed to be sterile. Its normal role is as an in-process control on the solution about to be sterile-filtered, where a typical action limit is 10 CFU per 100 mL of filter feed.
A bioburden count on a finished vial is therefore an odd thing to report. It is real data, and it is the wrong test for something presented as injectable-grade: it says the vial is not heavily contaminated, and explicitly does not say the vial is sterile. A pass is compatible with several live organisms per vial.
Sterility
The absence test. Either membrane filtration (dissolve the product, filter through 0.45 um, transfer the membrane to media) or direct inoculation. Two media, two temperatures: Fluid Thioglycollate Medium at 30 to 35 degrees for aerobes and anaerobes, Soybean-Casein Digest Medium at 20 to 25 degrees for fungi and aerobes. Fourteen days. Any growth in either is a failure pending investigation.
Two things about it that people do not expect. First, it requires a method suitability demonstration: you must show that the product itself does not inhibit growth, by spiking with a panel of challenge organisms and confirming they grow within three days (bacteria) or five (fungi) in the product's presence. Skipping this is how you produce a false pass on an antimicrobial product. Second, it is statistically very weak, which is the subject of a different question in this tag, but the headline is that testing 20 vials from a lot with 0.1% contamination detects it about 2% of the time.
Endotoxin
Measures a molecule, not an organism. LPS from the Gram-negative outer membrane persists long after the cell is dead and after any sterilising process. Its two blind spots: it sees nothing from Gram-positive bacteria or fungi, and it says nothing about whether anything is alive.
Now: "sterile filtered (0.22 um)"
What passing a solution through a 0.22 um membrane achieves: removal of organisms larger than the retention rating. The validation challenge organism is Brevundimonas diminuta at 10^7 per square centimetre, which is why 0.22 um is the accepted sterilising grade.
What it does not achieve:
- It does not remove endotoxin. LPS monomers are a few kilodaltons and pass a 0.22 um membrane as if it were not there. A solution can be sterile-filtered and grossly pyrogenic. This is the single most important thing in this answer.
- It does not remove mycoplasma or the smallest bacteria reliably; 0.1 um is used where that matters.
- It says nothing about what happened afterwards. The solution was filtered, then it was filled into vials, then partially stoppered, then loaded into a lyophiliser, then dried over a day or two, then stoppered and capped. Every one of those steps is an opportunity for ingress, and controlling them is the entire discipline of aseptic processing — laminar flow, environmental monitoring, gowning, media fills. A filtration step at the front of that chain guarantees nothing at the end of it.
- It is not a test. There is no result, no acceptance criterion, no method and nothing that could have failed.
And in a significant number of cases the phrase is simply inapplicable: if the vendor bought purified peptide powder, weighed it into vials and capped them, nothing was ever filtered. The line was copied from a template.
What a research-grade lyophilised peptide vial has actually been through
Realistically, for most of this market: synthesis, prep-HPLC purification, bulk lyophilisation of the pooled fractions, then weighing of powder into vials in a room that is clean but not classified, and capping. No terminal sterilisation, no aseptic fill validation, no sterility test, often no endotoxin test. Some larger houses do better and can show the data; assume the floor unless shown otherwise.
Which means the honest reading of your three COAs is: one has an endotoxin result, one has a bioburden count, and one has a sentence. None of the three has a sterility result. That is normal for the category, and the correct response is to know it rather than to read "sterile filtered" as reassurance. These are research-use-only materials not approved for human administration, and the documentation gap is one of several reasons that framing exists.
edited 23 Sept 2024 by seven_day_half — corrected a unit error in the worked example
3LPS passing a 0.22 um filter unimpeded is the fact that should be on the front page of every one of these COAs. – Dr_Marek_Zielinski 4 months ago 2The template-copying point is real. I have seen "sterile filtered" on a COA for a powder the vendor openly said they never dissolved. – Dr_Ingrid_Baumgartner 3 months ago Method suitability on the sterility test is the other silent failure. An antimicrobial product can pass by suppressing its own test. – v_ramaswamy 7 months ago add a comment