This is a 3-part endotoxin contamination story. The contamination part was easy. I cover that in this first part. But we pulled on a string that turned this into the craziest investigation I ever worked on. Now- Since this is an endotoxin story, I need to have the click bait picture used for every endotoxin article. Horseshoe crabs being harvested for their blue blood: If you’re reading this, you probably have some background on:
If you don’t have that background, check out this radiolab episode. It’s crazy how the horseshoe crabs help us. I’ll dive into the test they help us with in part 2. The only background you need to know today: some bacteria release endotoxin when they die, and it’s bad if it’s in our injectable drugs. Also- Endotoxin is measured in Endotoxin Units (EU). Onto the contamination story. We got repeat failing endotoxin results from our distillations units. We got them seemingly randomly once or twice a year. These distillations units were the last line of defense between city water and our Water For Injection (WFI) distribution system. Distillation units are the standard in pharma water purification. They essentially boil “dirty” water, collect the steam, and condense it into pure water. Sure enough, sampling and testing contamination was blamed for each failing result. I wasn’t in an investigator role for most of those results. But when I got my chance, I used it to learn about the entire pre-treatment system. It was set up like this: 1st- We start with incoming city water. We usually had low bioburden counts because of the chloride/fluoride combo in city water. City water pipes are covered in biofilms, so we always had endotoxin from the dead bacteria that flaked off. There’s no endotoxin limit for these city water samples, but testing is a compendial requirement. 2nd- Water softeners. Gotta remove those cations that can corrode piping! We didn’t do micro testing on this equipment. 3rd- Carbon Beds. These remove anions like the chloride and fluoride. Incidentally, they allow bioburden to grow rampant. With nothing to hold back bioburden growth, and plenty of carbon rich surface area to grow on, this is a bacteria paradise. Bioburden results were often higher than our tests could count (>250 CFU/mL). We didn’t do endotoxin testing here. That’s ok, because next we have… 4th- Those distillation units! The manufacturer claims a 6-log reduction in bacteria levels after water passes through them. They let 1 bacterium through for every 1 million coming in. They often get much better reduction than that. And, if the bacteria is gone, they can’t bring endotoxin into the system. The weird thing about this set-up: The site had two other pretreatment systems for connected buildings. They were designed pretty much the same, except they had RO filters between the carbon beds and the distillation units. RO filters cut down a lot of the bioburden and endotoxin. When the new system was designed, the engineers were so confident in the distillation units‘ power that it was considered a waste of money to put RO units in this new system. We knew distillation units were great at cutting down bioburden. But we didn’t have a bioburden issue. I went to the distillation unit vendor and got some documents like this. They only claim a 3-log endotoxin reduction. On a high endotoxin day city water had ~50 EU/mL. If we had a 3 log reduction, that puts us at 0.05 EU/mL, which was the limit of detection for the test (The specification is 0.25 EU/mL). When bioburden proliferated in the carbon beds, the endotoxin levels grew. We had to do a special protocol to confirm that. Turns out it’s much easier to kill off bioburden with steam than it is to denature the endotoxin they leave behind. Previous investigations didn’t look into this because:
Something interesting came from the city water data though. Results were consistently between 5 and 50 EU/mL. We always recovered some endotoxin, except for two days (out of years of data). A new lab manager asked why we wouldn't detect it those 2 days. I thought I knew why, so I looked into it to confirm. I was half right, but the reason I was half wrong is what makes this the craziest investigation I ever worked on. You’ll get to read all about that in Part 2 and Part 3! A final note on the decision for/against RO units before the distillation units- I made it seem like the site “cheaped” out by not using them. Industrial RO units are expensive! Given how infrequent the high endotoxin results were, how diluted That endotoxin became when it mixed with the rest of the water in the system, and how simple it was to wash the carbon beds, there’s a case that the costs/benefits work out in that systems favor. To address this issue, we increased the flushing of the carbon beds, which reduced the bioburden/endotoxin loads on them. The horseshoe crab picture is from this website.
0 Comments
Leave a Reply. |