If you make AAV or lentiviral vectors, you have almost certainly shipped batches that lost potency between the tank and the vial, and you may not have been able to say exactly where. The particle counts looked right. The genome titer by qPCR looked right. Yet the functional titer, the infectious or transducing units your patients actually depend on, came in low. This is the defining problem of viral vector downstream processing: the most damaging losses are invisible to the assays most teams run day to day, and they cluster around one component almost nobody specifies for.
Viral vector capsids are an order of magnitude more fragile than the monoclonal antibodies most tangential flow filtration and pump equipment was engineered for. AAV is a roughly 25 nm icosahedral protein shell. Lentivirus is a roughly 100 nm particle wrapped in a labile lipid envelope. Both lose function when their structure is disturbed, and much of the equipment in a standard downstream train inflicts that damage without ever registering on a routine yield calculation.
The measurement gap is the trap. A broken capsid still binds an antibody in an ELISA. A stripped lentiviral particle still shows its genome by qPCR. So analytical, physical titer (particles per mL) reads as intact while functional titer silently falls. The loss is often only detected at lot release or in clinical performance, when the batch is already gone and the run is unrepeatable at that cost.
The dominant loss point is not the membrane. It is the TFF UFDF recirculation pump. Peristaltic pumps, the default in most single-use trains, damage fragile vectors through three distinct mechanisms at once: high shear at the roller pinch point, micron-scale particle generation from tubing wear that both fouls the membrane and collides with capsids, and pressure pulsation that swings transmembrane pressure with every roller pass. Published viral vector literature attributes 20 to 40 percent functional titer loss to peristaltic-pumped TFF that never appears in particle counts.
The variable is the pump, not the membrane. For viral vectors the membrane is usually already validated. What silently determines your yield is the recirculation pump, and the standard peristaltic choice is the single largest source of invisible functional-titer loss.
Recirculation multiplies that damage. TFF passes the product through the pump hundreds of times over a multi-hour, and increasingly multi-day, step. The quantity that hurts your vector is not the peak shear of one pass; it is the cumulative dose, shear per pass multiplied by the number of passes. As continuous and semi-continuous processing lengthens run times, the same architectural choice compounds into steadily larger losses.
TFF carries three jobs for viral vectors, and it does them at two points where damage is most consequential. First it concentrates, removing bulk water from hundreds of liters of low-titer clarified harvest to raise titer. Then it performs buffer exchange and diafiltration. Finally it formulates the drug substance before fill-finish. It appears at capture concentration (after clarification, before chromatography) and again at final formulation (after polishing, before fill-finish). Any damage at that second point lands directly in the drug-substance specification. Typical AAV membrane MWCO is 100 to 300 kDa for a roughly 25 nm particle.
The failure modes differ by modality and have to be designed around individually rather than treated as one shear-sensitive category.
| Modality | Dominant failure mode | What TFF must protect |
|---|---|---|
| Lentivirus (~100 nm) | Lipid envelope ruptures under shear, foaming, air-liquid interfaces and pressure swings; inherently low titer with heavy recovery losses | Steady transmembrane pressure, closed air-free path, minimal cumulative shear |
| AAV (~25 nm) | Aggregation and fragmentation on repeated peristaltic passage; worsened full/empty capsid distribution under shear and interfacial stress | Low per-pass shear across hundreds of passes; capsid integrity and full/empty ratio |
| Adenovirus | Sturdier, but still erodes potency at scale over long runs | Cumulative-dose control on longer and continuous runs |
Beyond the pinch point there are further mechanical hazards. Cavitation on the pump suction side collapses vapor bubbles right at the product. Air-liquid interfaces and foaming denature and aggregate enveloped vectors. Concentration polarization at the membrane wall raises local stress. Membrane fouling drives transmembrane-pressure creep, which is worse on hollow fiber than on flat-sheet cassettes for lentivirus. Genome integrity and the full/empty capsid ratio are critical quality attributes that every one of these stresses can degrade.
There is also a containment dimension. Many nominally single-use TFF systems are single-use only on the membrane and tubing but reusable on the pump. That reintroduces cleaning-validation and cross-contamination exposure on the exact component doing the most damage, which is the last place a high-value, low-volume, closed viral vector process wants a reusable part.
TFFi is single-use tangential flow filtration built for the concentration, buffer exchange and formulation steps viral vectors need, at the 30 mL to 10 L batch volumes typical of AAV and lentiviral drug substance. It is built pump-first. Instead of recirculating on a peristaltic pump, it recirculates on the PIXER pump, an ultra-low-shear, near-pulseless positive-displacement single-use diaphragm pump, and holds transmembrane pressure stable. That targets the single biggest source of silent functional-titer loss directly. Because TFFi is membrane-agnostic, it works with the 100 to 300 kDa cassettes you have already validated for AAV. TFFi won the Interphex 2026 Best Technology Innovation award.
PIXER answers each peristaltic damage mechanism on its own terms. There is no roller pinch point, so there is no high-shear compression zone. The sealed single-use flow path generates no tubing-wear particles. Multi-diaphragm harmonic cancellation delivers near-pulseless flow that keeps transmembrane pressure steady for shear-sensitive lentiviral envelopes. Gravity-flooded suction eliminates the cavitation that damages capsids at the pump inlet. That combination is what protects functional titer across the hundreds of recirculation passes in a run.
Because PIXER packages the diaphragm and check valves as a single-use flow path, TFFi delivers a fully single-use recirculation loop: pump head, check valves, manifold, membrane and tubing. That removes the cleaning-validation risk and cross-contamination exposure that reusable-pump systems reintroduce on the most product-damaging component. TFFi runs on 24V DC with no compressed air, and VannX (motorized single-use diaphragm valve, plus/minus 0.3 PSI precision) plus ARTēVA single-use pinch valves extend the same electric, air-free, single-use control across the flow path. That supports the closed, air-free architecture that keeps air-liquid interfaces and foaming from denaturing and aggregating enveloped vectors.
PIXER handles viscosity up to 3,000 cP, which suits the high-concentration end of UFDF and final formulation, where concentrated viral vector feeds turn viscous and conventional pumps stall or spike shear. That is the exact point where damage lands directly in drug-substance specification. The Buffer Dilution System (inline, single-use, closed-system buffer management) supports the diafiltration stage of UFDF in the same closed, single-use manner, reducing open handling of fragile product.
All three major TFF vendors treat gentleness as a membrane choice or an operating parameter, and then default to a peristaltic recirculation pump. None owns the pump as the root cause. Alphinity names the variable: build the loop pump-first, and the 20 to 40 percent functional titer that peristaltic systems destroy invisibly is protected.
Because your assays measure the wrong thing for this failure mode. A broken AAV capsid still binds an antibody in an ELISA and a stripped lentiviral particle still shows genome by qPCR, so physical titer (particles per mL) reads as intact while functional or infectious titer silently drops. The damage is mechanical: shear, pulsation and cavitation from the recirculation pump degrade capsid integrity and rupture lipid envelopes without changing the particle count. You often only see the gap at lot release or in clinical performance, once the batch is gone.
A positive-displacement single-use diaphragm pump such as PIXER is gentler than a peristaltic pump because it has no roller pinch point, generates no tubing-wear particles, delivers near-pulseless flow and floods its suction by gravity to eliminate cavitation. Published viral vector literature attributes 20 to 40 percent functional titer loss to peristaltic-pumped TFF, a loss that never appears in particle counts. Over the hundreds of recirculation passes in a TFF run, the cumulative shear dose is what matters, so a lower per-pass shear compounds into a large yield difference.
Control the mechanical stress first, then the chemistry. Use a low-shear, near-pulseless recirculation pump to hold transmembrane pressure steady across the whole UFDF step, run a closed, air-free flow path so there are no air-liquid interfaces or foaming to denature enveloped vectors, and keep the loop fully single-use to avoid cleaning-validation carryover. The membrane MWCO for AAV is typically 100 to 300 kDa; that choice is usually already validated, so the variable left to protect is how gently the product is moved through the loop.
Specify the pump first. Pump architecture is the root cause of functional-titer loss, so start with a positive-displacement single-use pump, then its pulsation profile (near-pulseless keeps transmembrane pressure steady for shear-sensitive vectors), then confirm the flow path is fully single-use including the pump head to remove cleaning-validation risk, and only then match the membrane MWCO you have already validated. Membrane and MWCO matter, but they are not where most invisible loss originates.
How concentration, diafiltration and formulation come together in a single UFDF step, and where transmembrane pressure control decides your yield.
Why pump architecture, not membrane selection, sets the cumulative mechanical dose your product absorbs over a recirculation run.
Processing Viral Vectors & AAV?
Speak to an engineer