Bioprocessing Quarterly · Q2 2026
Three regulatory and research signals shaping single-use process decisions this quarter, and what risk-based extractables testing under USP <665> changes for fragile-modality manufacturers.
ASGCT 2026 wrapped in Boston this month with 2,100 abstracts and 172 sessions, and Q2 closes with three signals worth carrying forward into the next quarter. A peer-reviewed paper that gives the AAV single-pass TFF conversation real data. A fresh trade-press piece on the vendor data gaps that are slowing single-use adoption. And an extractables and leachables standard that just bound on 1 May. We end on what the third means for fragile-modality manufacturers.
Chaubal and colleagues published "Downstream Process Intensification for AAV Purification by Affinity Chromatography Using Single Pass Tangential Flow Filtration" in Biotechnology and Bioengineering earlier this year. The continuous-bioprocessing decision frameworks Andrew Sinclair and others have built at the strategic level now have a concrete tactical case to point at. The paper raises a question it does not fully answer: how pump architecture interacts with the SPTFF win on capsid integrity.
A featured Bioprocess Online piece this month surfaces what process teams have been saying privately for some time: the extractables, materials, and performance data suppliers publish is too thin or too generic to support real qualification under tightening standards. The story sits alongside NIIMBL's continuing work on data standardization and ontology for digital twins. The underlying conversation is the same: the vendor-side data layer is the bottleneck.
Risk-based extractables and leachables testing for single-use systems is now an operational requirement, not a guidance. Suppliers are publishing compliance briefs. Lab services firms are publishing testing menus. The question for manufacturers is what changes on the buyer side, especially for fragile-modality teams operating at scales the standard's frameworks were not written around. We spend time on this below.
USP <665> classifies single-use system components by contact risk: high, medium, low. Classification is driven by contact surface area, contact time, process conditions, and component composition. Testing requirements scale with the classification. On paper this is a sensible, proportionate framework. In practice it produces an asymmetry that fragile-modality manufacturers should be aware of.
A monoclonal antibody manufacturer running a 2,000-liter bioreactor with an established platform process and a well-characterized fluid path will typically classify lower-risk than an AAV manufacturer running a 50-liter perfusion process feeding multiple downstream concentration and diafiltration steps. Fragile-modality processes tend to feature longer cumulative contact times in unit operations like TFF, smaller batch volumes that raise the surface-area-to-product ratio for the same plastic, and process conditions (pH excursions, organic solvent diafiltration buffers, elevated temperatures) that push components into higher-risk classifications.
Two operational shifts follow. First, the supplier audit burden increases. Manufacturers cannot rely on a supplier's own classification under <665> because classification is process-specific, not component-specific. Asking "is this <665> compliant?" misses the point. The better question is: "what extractables data do you have under the temperatures, contact times, and buffer chemistries we actually run, and how was that data generated?" If the answer is a study at room temperature with WFI for thirty minutes, that data does not cover a buffer-conditioning step running at thirty-seven degrees for six hours.
Second, change control gets heavier. Substituting a tubing material, switching the vendor on a filter housing, or modifying a buffer chemistry now triggers a fresh risk evaluation under <665>. For ATMP teams already operating inside tighter CMC change windows than large-mAb shops, the friction compounds.
The patterns we are seeing across viral vector and cell-and-gene-therapy customers this quarter sort into three groups. The customers furthest along built the risk-evaluation framework first and then went to suppliers with data requests scoped to their own process. The customers still treating <665> as a supplier-side compliance question are running into requalification gaps. And a meaningful number of teams in the middle are discovering that components used routinely in mAb processes do not have the extractables coverage their fragile-modality processes actually require.
The piece of advice we keep giving and will keep giving: build the framework that fits your process first, then go demand data from suppliers that matches it.
The standard is the floor, not the ceiling. Build the framework that fits your process first, then go demand data from suppliers that matches it.