You are concentrating something an order of magnitude more fragile than the monoclonal antibody most bioprocessing equipment was built for, and the loss you fear most is the one you cannot see. An AAV capsid is a roughly 25 nm icosahedral protein shell. A lentiviral particle is around 100 nm wrapped in a fragile lipid envelope. CAR-T and allogeneic T cells are living material with a viability threshold. Each fails through its own weakest structure, and each carries a dose that is extraordinarily high in value and low in volume. When a batch is lost, it is not a commodity. It is patients not treated.
The hardest part is that the damage rarely announces itself. Mechanical stress can strip a lentiviral envelope or crack an AAV shell while the particle still registers in a physical titer assay, in particles per milliliter or by ddPCR. Functional and infectious titer drops, but analytical titer looks intact. The gap is often not caught until lot release or clinical performance, when the batch is already gone. This page is about where that loss happens and how to close it.
Every modality in this space has a different mechanical-damage threshold, and the equipment has to respect all of them. Lentivirus is the most shear-sensitive: its envelope disrupts quickly and infectious titer falls fast, and it is also sensitive to pH, salt and temperature. AAV is comparatively robust to shear but prone to aggregation, and it carries a heavy empty-capsid burden, often 50 to 90 percent of yield, that must be separated downstream. LNP and mRNA formulations shift their size distribution and can leak or expel payload under shear and interfacial stress. Adenovirus is sturdier but still erodes potency at scale. Living cells demand gentle, closed, automated washing and concentration to protect viability and function.
Process economics amplify every one of these losses. Separation and downstream operations can be roughly 70 percent of gene-therapy manufacturing cost. Because doses are so high in value and so low in volume, a few percentage points of yield or potency lost in a gentle-versus-harsh unit operation translate directly into cost of goods per dose and, ultimately, into how many patients a batch can serve.
Tangential flow filtration is the workhorse for concentration, buffer exchange and formulation, and it is also where the most damage accumulates. TFF shows up at two points in a viral-vector process: capture concentration after clarification and before chromatography, and final formulation or UFDF after polishing and before fill-finish. Because the retentate recirculates hundreds of times over a multi-hour step, the damaging quantity is the cumulative shear dose, shear per pass multiplied by the number of passes, not the peak shear of any single pass. Continuous and multi-day runs multiply that exposure further.
The hidden variable in functional-titer loss is usually the recirculation pump, not the membrane. Peristaltic pumps damage fragile product through three mechanisms at once: high shear at the roller pinch point, micron-scale particle shedding from compressed tubing that fouls membranes and collides with capsids, and pressure pulsation that swings transmembrane pressure on every roller pass, hitting enveloped and lipid-containing particles hardest. Peristaltic-pumped viral-vector TFF can lose a meaningful fraction of functional titer that never shows in particle counts.
The particle count is intact. The infectivity is not. That gap is the signature of mechanical damage, and it is written into the flow path long before it appears on a release assay. If you can only fix one variable in a fragile-modality TFF loop, fix the pump.
The loop holds other stressors that the pump sets in motion. Suction-side cavitation collapses vapor bubbles and releases destructive local energy right at the product. Entrained air creates air-liquid interfaces and foaming that denature and aggregate sensitive species. Concentration polarization piles product at the membrane wall, raising local shear and aggregation. Unstable transmembrane pressure makes all of these worse, and pulsation is what keeps transmembrane pressure unstable.
Alphinity's argument is simple and physical: the recirculation pump is the silent determinant of functional titer, and it should be engineered gentle rather than treated as a commodity. PIXER is a positive-displacement single-use diaphragm pump that attacks the number-one hidden cause of loss directly. Fluid moves through a chamber, not a roller pinch point or an impeller, so there is no deformation zone and no roller-shed particulate. Its multi-diaphragm odd-count harmonic cancellation produces near-pulseless flow, which holds transmembrane pressure steady and removes the per-beat pressure swings that hit lentiviral envelopes and LNPs hardest. Its gravity-flooded suction eliminates cavitation at the inlet, closing a source of particle damage that standard suction-fed pumps leave open. PIXER also handles viscosity up to 3,000 cP, so concentrated retentate late in a run does not force a harsher regime.
TFFi is the single-use tangential flow filtration system built on PIXER for exactly this modality: concentration, buffer exchange and final formulation of viral vectors from 30 mL up to 10 L. That range spans development through clinical batch sizes, so a gentle process characterized at small scale carries forward without changing the fluid mechanics the product sees. TFFi holds stable transmembrane pressure and is membrane-agnostic, so you keep the hollow-fiber or cassette membrane you already trust and fix the variable that actually determines yield, rather than re-qualifying separation media. It is GMP-compatible, runs on 24V DC with no compressed air, and won the Interphex 2026 Best Technology Innovation award.
| TFF loop variable | Conventional peristaltic loop | Alphinity PIXER / TFFi |
|---|---|---|
| Shear source | Roller pinch point, high local shear | Chamber displacement, no pinch point, ultra-low shear |
| Pressure profile | Pulsation swings TMP every roller pass | Near-pulseless, stable transmembrane pressure |
| Particle shedding | Micron-scale shed from compressed tubing | No compressed-tubing shed at the pump |
| Inlet cavitation | Possible on suction-fed inlet | Gravity-flooded suction, cavitation eliminated |
| Single-use scope | Often membrane and tubing only; pump head reused | Fully single-use including pump head and check valves |
| Scale continuity | Fluid mechanics can shift with scale | Same mechanics 30 mL to 10 L |
Closed, single-use, cleaning-validation-free flow paths are effectively mandatory for viral and cell products because of cross-contamination and bioburden risk. Yet many systems marketed as single-use are single-use only on the membrane and tubing while reusing the pump head, leaving a carryover bottleneck at the most product-contacting, highest-shear component. The PIXER flow path in TFFi is fully single-use including the pump head and check valves, so the closed, cleaning-validation-free path covers the whole product-contact surface.
Gentle flow control then extends beyond the filtration step to the rest of the skid. VannX motorized single-use diaphragm valves deliver plus/minus 0.3 PSI precision on 24V DC with no compressed air, and ARTēVA single-use pinch valves keep the flow path closed and single-use throughout. The Buffer Dilution System provides inline, closed, single-use management of the diafiltration and formulation buffers that UFDF steps consume in volume, keeping buffer prep inside the same closed single-use architecture. The result is one electric, air-free, single-use flow path for a fragile, high-value process from end to end.
When particle count (particles/mL or ddPCR) looks intact but infectivity is down, the loss is almost always mechanical damage to the particle, not the separation. Membrane fouling shows up as declining flux and rising transmembrane pressure over the run, and it retains material you can often recover. Pump and shear damage shows up differently: functional and infectious titer fall while physical titer holds, because a stripped lentiviral envelope or a cracked AAV shell still registers in a physical assay. The recirculation pump is the usual hidden variable, since the retentate passes through it hundreds of times over a multi-hour step. A practical test is to hold membrane, buffer and flux constant and change only the pump architecture. If functional titer recovers, the pump was the culprit.
Specify a positive-displacement single-use diaphragm pump rather than a peristaltic pump. Peristaltic pumps combine three damage mechanisms: high shear at the roller pinch point, micron-scale particle shedding from compressed tubing, and pressure pulsation that swings transmembrane pressure on every roller pass. PIXER moves fluid through a chamber rather than a deformation zone, so there is no pinch point and no roller-shed particulate, and its multi-diaphragm odd-count harmonic cancellation gives near-pulseless flow. Ask the supplier to prove the pulsation and transmembrane-pressure profile at your actual operating flow rate and back pressure, not at a convenient benchmark.
The damaging quantity is the cumulative shear dose (shear per pass multiplied by number of passes), not the peak shear of any single pass. Lowering crossflow to reduce per-pass shear can lengthen the step and raise total exposure, which is the shear-versus-flow-rate tradeoff. The way out is to lower the shear intensity the product sees on every pass at the source, the pump, so you can keep a workable crossflow and finish in a practical time. A gentle, near-pulseless pump lets you hold a productive flow rate while keeping cumulative dose low, which is exactly the regime lentivirus needs.
Yes. Many systems marketed as single-use are single-use only on the membrane and tubing while reusing the pump head, which leaves a cleaning-validation and carryover bottleneck at the most product-contacting, highest-shear component. The PIXER flowpath in TFFi is fully single-use including the pump head and check valves, so the closed, cleaning-validation-free path extends across the whole product-contact surface. Paired with VannX single-use diaphragm valves and ARTēVA single-use pinch valves, the entire skid stays closed, single-use and air-free, running on 24V DC with no compressed air.
How TFF concentration and buffer exchange work, and why cumulative shear dose, not peak shear, decides how much fragile product survives.
Why the recirculation pump, not the membrane, is often the silent determinant of functional titer in viral-vector and cell processing.
Processing Cell & Gene Therapy?
Speak to an engineer