Failures in CAR T-cell products manufacturing

by Alexey Bersenev on March 12, 2016 · 2 comments

in cell product

Post to Twitter Send Gmail Post to LinkedIn

One of the best events that I’ve attended recently was the first education (CME) course Clinical Application of CAR T Cells at Memorial Sloan Kettering Cancer Center (MSKCC). It was great event with stellar speakers from all major academic centers. Today, I’d like to touch a topic of manufacturing feasibility and failures in CAR T-cell production. The data, that I’m going to discuss here, were presented at MSKCC course as well as at ASH 2015 meeting.

Feasibility of CAR T-cell products manufacturing in hematology/oncology
Here are some reported data:

  • NCI: manufacturing feasibility 96% (failure rate 4%) in pediatric ALL, reported by Daniel Lee and Crystal Mackall
  • Upenn: 7% failure rate in CLL (1/14), reported by reported by David Porter
  • Upenn: 14% failure rate in lymphomas (6/43), reported by Steven Schuster
  • MSKCC: 6% failure rate (1/16) in adult B-ALL , reported by Marco Davila
  • MSKCC: Overall feasibility of production all type of CAR T-cell products 97.5%, reported by Isabelle Riviere
  • Fred Hutch: 8% failure rate (3/39) in lymphomas, reported by Cameron Turtle
  • Seattle Children’s: 40/41 products were released (2% failure rate) in pediatric B-ALL, reported by Rebecca Gardner

Causes of manufacturing failures
Based on what I’ve learned from the meetings, number one reason of manufacturing failure is inability to achieve targeted dose. The causes of this failure are the following:
– low number of T-cells in incoming apheresis product, measured as absolute lymphocyte count (ALC)
– poor selection of T-cells on day 0 of manufacturing
– irreversibly impaired T-cells (no response to stimulation in culture).

Based on MSKCC experience (reported by Isabelle Riviere), ~2.5% products failed due to inability to achieve targeted dose in culture. If targeted dose is not achieved, usually products get released anyway, and, usually patients respond to it (see Davila’s report as example). Daniel Lee (NCI) mentioned that 2 pediatric B-ALL patients responded well to lower dose (“failed”) products.

As example of failure due to poor T-cell selection, contamination of cell culture by monocytes/ granulocytes was mentioned by Daniel Lee and Crystal Mackall (NCI). Monocytes removal can improve feasibility of CAR T-cell product manufacturing and also can rescue previously failed attempt to expand T-cells. At least 2 products were rescued by monocyte depletion in pediatric ALL trial at NCI.

Low T-cell number (measured by ALC) was cited by Steven Schuster as a major reason for relatively high (14%) manufacturing failure in Upenn’s lymphoma trial. He said that low ALC is very typical for heavily pre-treated lymphoma patients, enrolled in the trial (sometimes low ALC obviate a need for pre-CART therapy lymphodepletion). What is considered low number of T-cells in blood and in the apheresis product? Schuster said many lymphoma patients had ALC < 200 cells/uL. The case of manufacturing failure from published MSKCC study, was associated with 3.7% of T-cells in apheresis collection.

All causes of potential manufacturing failures, mentioned above, are specific for T-cell products. However, other, more general causes, such as microbial contamination, equipment-related cell loss, high endotoxin level, accidents and other, can make CART-cell products unsuitable for release.

Potential solutions
I’d highlight two major ways for prevention of manufacturing failures: stringent inclusion criteria (patients selection) and improvement of cell processing.

    1. Setting specs for apheresis collection. I was discuss this option here. As an example, you can look at inclusion criteria for CD19-CAR T-cell pediatric B-ALL trial at Seattle Children’s – ALC must be > 100 cells/ uL.
    2. More stringent inclusion criteria. If patients were heavily pretreated by drugs, affecting T-cells, they may be excluded.
    3. Removal of accessory cells from apheresis product. One of the best examples of this option is monocytes depletion by adherence, practiced at NCI and MSKCC. Implementation of monocyte depletion at NCI allowed to increase manufacturing success rate from 86% (4/28) to 96% (see fragment of Crystal Mackall’s ASH 2015 presentation here). The same approach is currently used at MSKCC. As per Isabelle Riviere (MSKCC), about 2.5% of products need CD14 depletion for successful manufacturing.
    4. Better purification of T-cells before culture. Ideally, T-cell subpopulation could be sorted (not just enriched) from apheresis collection with purity >95%. Fred Hutchinson Cancer Center is pursuing this approach with concurrent CD4 and CD8 sorting and generation of products with well defined composition.
    5. Changing “targeted dose” in product’s release criteria.

{ 2 comments… read them below or add one }

Ron Smith March 15, 2016 at 5:53 am

You are missing one big part of the story. NVS/UPenn do a “test expansion” prior to accepting a patient into a study. Depending on the patient indication, this can be 25-40% of potential patients.

Their manufacturing success rates are only for patients accepted for manufacturing.


Alexey Bersenev March 15, 2016 at 10:13 pm

Thank you, Ron! Good point.
So, if you remove test expansion the manufacturing failure rate will go up.


Leave a Comment

Previous post:

Next post: