Photo of a clinical professional reaching into a BSC to culture cell therapies.
Today’s most popular commercial separation approaches fail to capture sufficient numbers of cells with a high therapeutic value.
iStock, Blackholy

Solving Cell Therapy’s “Garbage in, Garbage out” Problem with Automation

Despite industry’s emphasis on cell engineering and expansion, better starting materials are the foundation of the manufacturing revolution

Tony Ward

Tony Ward is chief technology officer at Curate Biosciences. Curate is developing its Cell Processing System, which applies microfluidics-based deterministic cell separation (DCS) technology for T cell isolation through a...

ViewFull Profile
Published:Jul 07, 2022
|4 min read
Photo portrait of Tony Ward
Tony Ward

Tony Ward is chief technology officer at Curate Biosciences. Curate is developing its Cell Processing System, which applies microfluidics-based deterministic cell separation (DCS) technology for T cell isolation through a benchtop system and single-use cartridges. Tony was previously global strategic marketing director for Becton Dickinson’s research cell analysis business, headed commercial operations for eBioscience, and served as SVP/general manager for Affymetrix’s business unit.


With thousands of successful treatments, thousands of candidates in the pipeline, and novel designs in testing in every disease area, cell therapy is clearly booming. The spectacular success of several autologous CAR-T cell therapies—including remission rates of up to 93 percent against some of the most aggressive hematologic cancers—have helped spur almost $24 billion in global investments over the past three years. Yet it has an Achilles’ heel: the starting material is of variable, and often of poor, quality.

Cell separation is a key initial step for any autologous therapy, isolating the starting materials to later be engineered into a therapy. This not only means separating immune cells from much more populous red blood cells, but from factors like platelets that can induce naive T cells to differentiate into unsuitable and/or mature subtypes.

Today’s most popular commercial separation approaches fail to capture sufficient numbers of cells with a high therapeutic value, forcing drug developers to perform a series of time-consuming engineering steps that can also negatively impact efficacy. The goal instead should be to capture the maximum number of naive immune cells possible in the initial separation process.

Unspinning a complex web

Current methods for cell separation in autologous therapy manufacturing are limited in different ways. One approach is density gradient centrifugation with a Ficoll solution. This method is relatively effective in eliminating most red cells, which outnumber lymphocytes 1,000:1 in whole blood. But it requires numerous wash cycles to remove platelets, resulting in significant cell loss. All told, the separation process can take more than four hours to perform manually and can result in the loss of up to 80 percent of the potential therapeutic cells. As a result, centrifugation with manual Ficoll steps is increasingly rare, and today is typically only used to support development and Phase 1 work.

A more common approach is counterflow centrifugal elutriation, which can be used to perform a gross fractionation of leucocytes from apheresis. The elutriation process has to be combined with a cell loss inducing lysis step to remove red cells. As a result, maximizing recovery and purity requires skilled and expert processing, and still results in cell loss.

The third common separation method is magnetic selection, which recovers only about 60 percent of abundant target cells, and as few as 20 percent for rare cells, such as stem cells. Cells processed with magnetic separation alone also have a less consistent cell expansion profile, forcing users to combine this strategy with other complex processing steps.

Reaping what you sow

The limitations of commercial cell separation approaches impact more than just the first step of manufacturing. It’s analogous to what data scientists working with artificial intelligence refer to as the “garbage in, garbage out” problem—where the poor quality of inputs lowers the ceiling for possible outputs.

First, there’s the labor bottleneck. There’s no single uniform strategy to produce enough quality cells for the many different types of cell therapies. The result is highly manual workflows cobbled together for each bespoke process, demanding specialized expertise and large spaces that can accommodate multiple independent systems. That’s typically most efficient at a centralized location, squandering precious time shipping materials from apheresis site to processing site.

Although a growing body of research suggests that prolonged cell culture and expansion time affects the quality of an autologous therapeutic, today, a variety of time-consuming engineering steps are required to expand an isolated T cell population to sufficient numbers. This highly manual, specialized, labor-intensive, and human error-prone process can take two weeks—by far the longest part of autologous cell therapy manufacturing.

The production and labor bottlenecks are intensely important because of the target patient population. Approved autologous therapies are largely reserved for patients with late-stage hematological cancers who have failed all other courses of treatment and can least afford protracted waits. Also troubling: these patients often have a weakened immune system that cannot produce enough quality starter T cells at apheresis, given the percentage that will be lost during separation. This leads to an estimated 20 percent of patients that die before they can receive CAR-T treatments because of long production times.

These manufacturing challenges are also a major reason why cell therapies are notoriously difficult to scale up—and exorbitantly expensive. Today, they can cost as much as $450,000 per patient. Producing doses more quickly and in a more cost-effective manner is thus key to the future of cell therapy.

Automation revolution

Automation is the next big step for autologous manufacturing, with the potential to address multiple bottlenecks. In cell separation, the most promising approach may be enclosed microfluidics systems, which allow for rapid size-based sorting of cells without chemical or mechanical stressors. Recent data suggests that such systems can help recover a much larger percentage of key immune cells including T, B, and NK cells in a fraction of the time and at the site of apheresis. And by isolating more of the cells before they are degraded or lost, they are more likely to differentiate into efficacious therapeutics.

If manufacturers can find a way to scale autologous cell therapies to meet burgeoning demand, it has the potential to reshape the health care system, delivering potentially curative therapies for a series of intractable, chronic, and frequently deadly illnesses. The best place to start is the beginning—automating the extraction of the starting materials. By decreasing or obviating the need for subsequent cell expansion, manufacturers can quickly return more effective autologous therapies to patients in desperate need.