White Light Spectroscopy for Mammalian Cell Viability/Quality

Assessment: Towards an Online, Label-Free and Sampling-Less

System to Simplify Quality Control in CAR T-Cells Production

Bruno Wacogne

1,2

, Céline Codjiova

, Jovanne Palvair

, Naïs Vaccari

, Mélanie Couturier

Alain Rouleau

and Annie Frelet-Barrand

Université de Franche-Comté, CNRS, institut FEMTO-ST, F-25000 Besançon, France

Centre Hospitalier Universitaire de Besançon, Centre d’Investigation Clinique, INSERM CIC 1431,

25000, Besançon, France

MedInnPharma, 4 rue Charles Bried, 25000 Besançon France

Keywords: White Light Spectroscopy, Mammalian Cells, Viability, Car T-Cells.

Abstract: CAR T-cells are highly promising medical products for personalized medicine, but their long production is

complex and require extensive quality controls, which result in prohibitive costs for most patients. These

controls include monitoring of cell concentration, viability, and possible contamination detection. To simplify

CAR T production, these controls should ideally be conducted online in a closed system, without sampling

from the bioreactor. Recently, we proposed white light spectroscopy as a method for online monitoring of cell

concentration. In this paper, we demonstrate that this optical method can also assess cell viability. We define

a cell suspension "quality value" which shows a linear relationship with cell viability estimated by

conventional methods. This relationship varies depending on the techniques used and the dominant T-cell

death process induced. We then hypothesise that the quality score could serve as a general indicator of cell

suspension health, as it is not dependent on any biophysical-chemical interaction or instrument. Overall, the

correlation between conventional and optical methods, together with previously published results on cell

concentration monitoring, suggests that white light spectroscopy is a promising on-line and sample-free option

for monitoring CAR T production.

1 INTRODUCTION

CAR T-cells are promising advanced therapy drugs.

Their manufacturing process is quite complex (Wang,

2016; Wang, 2023) and quality control is daily

performed at each step and especially during the

expansion phase. Quality control reside mostly in

monitoring cell concentrations, assessing cell

viability, and detecting potential contamination. We

have recently proposed a proof of concept for a

potentially sampling-less white light spectroscopy

system for monitoring T-cell concentrations

(Wacogne 2020, 2021, 2022). The goal of our current

work is to explore the potential of white light

spectroscopy for assessing cell viability. Indeed, cell

viability measurement is a critical technique in

biological research, drug development, and clinical

applications used to assess cell health, proliferation,

and survival under various experimental conditions.

However, cell viability assessment remains a

challenge for label-free and online applications. A

wide range of methods have been developed to

quantify/measure cell viability, based on specific cell

type, experimental objectives, and available

resources.

Traditional Techniques

The trypan blue dye exclusion assay is the historical

technique of assessing cell viability (Macklin, 1920).

It distinguishes live cells from dead ones based on

membrane integrity. Viable cells exclude the dye,

while dead cells retain it (Louis, 2011). In some cases,

trypan blue may also have an effect on cell

morphology and viability depending on dye

concentration (Tsaouis, 2013). Cell sampling is

required as cells must be collected from the culture,

Wacogne, B., Codjiova, C., Palvair, J., Vaccari, N., Couturier, M., Rouleau, A. and Frelet-Barrand, A.

White Light Spectroscopy for Mammalian Cell Viability/Quality Assessment: Towards an Online, Label-Free and Sampling-Less System to Simplify Quality Control in CAR T-Cells Production.

DOI: 10.5220/0013149400003911

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 18th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2025) - Volume 1, pages 53-64

ISBN: 978-989-758-731-3; ISSN: 2184-4305

mixed with the dye, and then manually counted using

a hemocytometer or automated cell counter. Although

cost-effective, its low sensitivity and manual counting

introduce limitations, particularly for high-

throughput or highly accurate experiments (Stoddart,

2011). However, the use of Ni

or Co

salts may

help improving the assay sensitivity (Sarma, 2000).

Other techniques could be used. The MTT and

MTS assays measure mitochondrial reduction of

tetrazolium salts into formazan by viable cells. Cells

do not need to be sampled from the culture for

immediate analysis but, at the end of the assay, the

cell medium is sampled to measure the color change

associated with formazan production using a

spectrophotometer. This method, though easy to

perform, depends on mitochondrial activity, which

can vary with cell type and condition (Berridge,

2016). In the case of irradiated HepG2 cells (Chung,

2015), a decrease of cell viability was observed using

trypan blue while no significant changes could be

detected using MTT assay. This indicates that the

viability detection principle should be adapted to the

cell death process.

LDH assays can also quantify cell death by

measuring LDH released by damaged cells into the

medium (Zou, 2013). Sampling of cell supernatant is

required to analyse extracellular LDH levels,

typically performed after an incubation period to

allow sufficient LDH release. While this method is

sensitive to early cell death, it cannot distinguish

between apoptosis and necrosis. LDH and MTT

assays were compared to study cytotoxicity of

cadmium chloride on HTC and HepG2 cells (Fotakis,

2006) and MTT assay exhibit a higher sensitivity

compared to LDH assay.

Advanced Techniques

Other methods have been developed. Among them,

flow cytometry enables precise identification of live,

apoptotic and dead cells by using reagents such as

fluorescent dyes, cell impermeant viability dyes or

leukocyte markers for example. Sampling of cell

suspension is required to collect and stain cells before

analysis through cytometer. This technique is a gold

standard for high-throughput applications due to its

precision in cell viability measurement (Shenkin,

2007) and allows detecting both apoptosis and

necrosis (Kumar, 2015).

Fluorescence microscopy assays use different

dyes to stain live cells (for example, AlamarBlue;

Hamalainen-Laanaya, 2012). Dyes are added to cell

culture, and after incubation, images are captured to

quantify cell viability. The accuracy of automated

image analysis and high-content screening provides

quantitative results, especially for high-throughput

studies. However, exposure to excitation light may

alter cell viability itself and tolerable light doses must

be employed (Schneckenburger, 2012). It was found

that the cell morphologic changes due to trypan blue

makes them difficult to count under microscope

generating an artificially higher viability compared to

fluorescence methods and could result in viability

measurement differences between these methods

(Chan, 2015).

Impedance-based methods can track cell viability

by measuring electrical impedance as cells attach and

spread on electrode-coated plates. It is label-free,

non-invasive and monitors cell continuously and in

real time, offering high sensitivity and accuracy

without disrupting the cells and does not require

sampling of cell solution. However, cells must spread

on coated plates or micromachined substrates which

makes it difficult to implement in conventional

laboratories (Optiz, 2019; Zhong, 2021; Yang, 2023).

Microfluidic devices enable high-throughput and

low-volume viability assays at single-cell level.

Sampling is not typically required as the fluidics

system controls cell and reagents introduction and

analysis is performed within microchannels. Cell

viability assay based on image processing of stained

cells was proposed for microfluidic 3D culture (Ong,

2020).

Commercially Available Systems

Several commercial systems for cell viability

measurement are available on the market, offering

various features in terms of sensitivity, throughput

and accuracy. Below are some of the most widely

used systems:

• The xCELLigence Real-Time Cell Analysis

(RTCA) System from ACEA Biosciences (Agilent) is

based on label-free impedance measurements. It

supports high-throughput applications with 96-well

plates but can hardly be implemented for online

measurements.

• The LUNA II (Logos Biosystems) and Vi-

CELLTM XR Cell Viability Analyzer (Beckman)

automate the trypan blue exclusion assay after

sampling of cell suspension.

• The Fluidlab-R-300 (System C Bioprocess) rely

on lens less imaging or holographic digital

microscopy. It also allows optical spectral recording.

• Specific kits can be purchased for

viability/cytotoxicity assessment. The

LIVE/DEAD™ Viability/Cytotoxicity Kit (Thermo

Fisher) allows fluorescence measurement which can

be measured by cytometers, fluorescence microscope

and automated plate readers.

BIODEVICES 2025 - 18th International Conference on Biomedical Electronics and Devices

In summary, the measurement of cell viability has

evolved significantly with methods ranging from dye

exclusion assays to advanced real-time label-free

technologies. Traditional techniques remain widely

used because of their simplicity, while new methods

such as impedance-based assays or microfluidics offer

improved accuracy and physiological relevance. The

choice of method depends on specific experimental

context, balancing cost, complexity and/or data

quality. It should be noted that the discrepancy between

methods is related to either predominant T-cell death

process and/or type of biophysical interaction

phenomena exploited for viability measurement. Most

of the methods presented above require sampling,

while others rely on mixing reagents with cell culture.

Therefore, they are not suitable for label-free, online

and/or real-time measurements.

In this paper, we propose a proof of concept for a

white light spectroscopy method to assess cell

quality/viability without specific reagents using

lymphocyte line and inducing cell death processes.

The method can be easily transferred to an online and

sampling less system. The materials and methods are

described in section 2, while the results are presented

in section 3. A discussion is then proposed including

some medico-economic considerations about CAR T-

cells before concluding.

2 MATERIAL AND METHODS

The CEM-C1 T lymphoblast line was used in this

study. Cell death was induced in two ways: X ray

irradiation and storage at a temperature between 4°C

and 37°C. Cell viability of both normal and

dying/dead cells was then measured using an

automated cell counting, flow cytometry, and white

light spectroscopy. The aim is to establish a

spectroscopic criterion that accounts for the loss of

cell integrity, and to correlate this criterion with

viabilities measured by conventional techniques.

2.1 Design of the Experimental

Protocols

X-Ray Induced Cell Death

A total of 7 experiments were performed with both

normal and irradiated cells. The experimental

protocol is shown in figure 1(a).

Low Temperature -Induced Cell Death

A total of 3 experiments were performed with cells

stored at 4°C, 25 °C and 37°C for each experiment.

The protocol is presented in figure 1(b).

Figure 1: Design of the experimental protocols and spectra

measurement setup. (a) Cell death is induced by (a) X-ray

exposure (n = 7). (b) storage at low temperature (n = 3). (c)

Absorption spectra measurement setup (from (Wacogne,

2022)).

2.2 Cell Culture, Death Induction and

Viability Measurements

Cell Culture

The human T lymphoblast T-cell line CEM-C1

(ATCC® CRL-2265 TM) was cultured between

5×10

and 2×10

cells/mL in RPMI 1640 medium

without Phenol Red (Gibco #11835030) and

White Light Spectroscopy for Mammalian Cell Viability/Quality Assessment: Towards an Online, Label-Free and Sampling-Less System to

Simplify Quality Control in CAR T-Cells Production

supplemented with 10% FBS (Gibco #A5670701)

and 1% penicillin/streptomycin solution (PS, Gibco

#15140122) at 37 ℃ in a 5% CO

humidified

incubator. CEM-C1 were seeding at 6×10

cells/mL

in RPMI 1640 medium without Phenol + 10% FBS +

1% PS.

Cell Death Induction

For X-ray death induction, suspensions were

submitted to X-ray exposition for a total dose of 35Gy

for 7 min. Non-irradiated and irradiated cells were

then cultured at 37 ℃ in a 5% CO

humidified

incubator and monitored during 48h.

For low temperature death induction, suspensions

were simply stored at 4°C, 25°C and at 37°C. They

were monitored during 24 hours.

Viability and Quality Measurements

For X-ray death induction, cell viability was assessed

in three ways depending on equipment availability. (i)

A LUNA-II automated cell counter (Automated Cell

Counter, Logos Biosystems) was used on 5/7

experiments by mixing (V/V) cell suspensions and

Trypan Blue (15250061, Fischer Scientific). (ii)

Viability, apoptosis and necrosis were evaluated by

cell staining using Annexin-V coupled to FITC and

7-AAD according to provider’s instructions (Annexin

V Apoptosis Detection Kit, BD Bioscience #556547).

Viable (AnV-7AAD-), apoptotic (AnV+7AAD-) and

necrotic (AnV+7AAD+) cell percentages were

determined by flow cytometry (SP6800, Sony

Biotechnologies). (iii) White light spectroscopy was

used as detailed below and in section 2.3 on the 7

experiments.

For low temperature death induction, only

automated counting and spectroscopy were used.

2.3 Spectra Acquisition Setup and Data

Processing

The experimental setup for spectra acquisition is

simple as it only includes a white light source, a

cuvette holder and a compact spectrometer (figure

1(c) issued from (Wacogne, 2022)). Spectra were

acquired with 3647 data points between 177 nm and

892 nm wavelength. They were truncated to keep the

range between 350 and 850 nm wavelength where the

signal-to-noise ratio is higher.

Spectra were recorded in transmission and

converted to absorption spectra for mathematical

treatments. Trends of experimental data were

calculated suing the “Smoothing Spline” feature of

the Matlab

Curve Fitting toolbox with a smoothing

parameter set to 2.25×10

-3

Spurious peaks due to strong emission lines from

the deuterium lamp (around 485 and 655 nm

wavelength) were mathematically removed, as well

as the still not fully understood additional signal at

410 nm wavelength (Wacogne, 2023). Pre-processing

and subsequent data processing were performed with

Matlab

version R2020b.

For X-ray death induction, transmission spectra

were recorded every 2 hours on the first day and at

T=24, 30, and 48 hours thereafter. Some spectra were

missing due to experimental difficulties: T0 for

normal and irradiated cells exp. #4 and #5, T8 for

normal and irradiated cells exp. #6. 3 normal cell

spectra were not included due to too high cell

concentration resulting in saturated absorption

spectra: T48 exp. #2, T30 and T48 exp. 3. In total, 50

spectra were recorded with normal cells and 53 with

irradiated cells. They were processed to calculate

quality values (see below), which were compared

with viabilities measured by either the LUNA-II

automated cell counter or by cytometry.

For low temperature death induction,

experiments were performed in two ways: (i) cells

were prepared in the morning, stored at different

temperatures and transmission spectra were

recorded every 2 hours until the morning after

(except during the night), (ii) cells were prepared in

the evening, 1 transmission spectrum was recorded

in the evening and the others the next day. A total of

72 measurements were made.

3 RESULTS

3.1 Defining a Quality Value of Cell

Suspensions

The basic principle is based on a number of

observations. (i) Healthy cells divide efficiently and

few small particles (vesicles, apoptotic and necrotic

bodies…) are generated, leading to high viability.

Conversely, a culture with low cell viability contains

a high concentration of small particles. (ii) From an

optical point of view, a suspension of healthy cells

contains mainly cells (large particles), light

propagates mainly in a straight line, and light-matter

interaction results mainly in absorption. (iii) A low-

quality suspension contains a large number of small

particles that scatter light. This scattering increases

according to 1/λ for micron-sized particles. From this

we deduce that a high viability culture will have an

absorption spectrum that reflects only the absorption

of the cells, and whose shape is vaguely Gaussian,

figure 2(a). Conversely, a low viability culture will

BIODEVICES 2025 - 18th International Conference on Biomedical Electronics and Devices

see the shape of its absorption spectrum distorted by

the 1/λ component of the scattering, figure 2(b). Thus,

the higher the quality of a culture, the more Gaussian

the shape of its absorption spectrum. Note that the

spectra shown in figure 2 and 3 correspond to

unpublished data and are used for illustrative

purposes.

Figure 2: Examples of cell suspensions with (a) high and

(b) low viability.

It must therefore be possible to estimate the

viability of a suspension by the more or less Gaussian

appearance of its absorption spectrum. One method is

to fit the absorption spectrum with a Gaussian

function. The R

of the fit then reflects the Gaussian

aspect of the spectrum and is therefore a measure of

the quality of the cell suspension. A suspension

quality value is defined by: Q=100*R

expressed in

%, R

resulting from the fitting of the absorption

spectrum by the following function:

𝑎𝑏𝑠



𝜆



𝑎.𝑒𝑥𝑝

𝜆𝑏

𝑐





𝑑

)

In equation (1), 𝑎𝑏𝑠



𝜆



is the absorption

spectrum, a, b, c and d are the fitted coefficients.

Figure 3 shows examples of fitting obtained using

equation (1) on both types of suspensions.

Figure 3: Examples of Gaussian fitting with (a) high and (b)

low viability.

However, a few precautions need to be taken. If

no constraints are imposed on the fitting coefficients,

the algorithm always manages to find a quadruplet

leading to a high R

. The constraints imposed on the

coefficients in this work are shown in Table 1.

Table 1: Variation intervals of fitting coefficients.

Coeff.

(

)

(

)

(

)

(

)

Low. lim.

0 550 0 0

Upp. Lim.

200 700 500 200

3.2 Comparison of Viability and

Quality of Non-Irradiated and

Irradiated Cells

Figure 4 shows examples of spectra recorded with

non-irradiated and irradiated cells. For non-irradiated

cells, the concentration increased from 6×10

13×10

cell.mL

-1

during the 48-hours experiment.

The generation time was 44 hours, which was longer

than that measured in previous experiments due to

slight differences in the culture protocol.

Concentrations were measured by the method

described earlier (Wacogne, 2022). For irradiated

cells, the concentration decreased from 6×10

2×10

cell.mL

-1

, which was expected. In this case, the

concentration was measured spectrally using a more

advanced method than the one used in 2022.

White Light Spectroscopy for Mammalian Cell Viability/Quality Assessment: Towards an Online, Label-Free and Sampling-Less System to

Simplify Quality Control in CAR T-Cells Production

Figure 4: Examples of spectra recorded with (a) non-

irradiated cells and (b) irradiated cells. Green arrows show

either increasing or decreasing concentration.

Figure 5 shows quality values calculated from

absorption spectra and viabilities measured with the

automated counter and the cytometer, along with the

corresponding trends corresponding to the data

shown in figure 4.

Figure 5: Quality values and viabilities corresponding to

spectra of figure 4 for (a) non-irradiated cells and (b)

irradiated cells. Markers: experimental data. Lines: trends.

A correlation was observed between qualities

measured with white light spectroscopy and

conventionally measured viability. Trends show an

unexpected increase in viability as measured by

cytometry at the end of the experiment. This was

often observed and will be addressed in section 4.

Experimental data show dispersion around

corresponding trends. Dispersion was calculated

using a modified form of the standard deviation

calculation where the mean is replaced by the trend as

follows.

𝑖𝑠𝑝







∑

𝑉𝑎𝑙𝑢𝑒



𝑇𝑟𝑒𝑛𝑑











)

In equation (2), Value

represents the i

data point

and Trend

is the value of the corresponding trend at

data i.

Dispersion was then calculated for each

measurement method for both non-irradiated and

irradiated cells (figure 6).

Figure 6: Boxplot of dispersion for the different

measurement methods for (a) non-irradiated and (b)

irradiated cells.

Dispersion was quite low for non-irradiated cells

(median less than 3%), while it was significantly

higher for irradiated cells (median over 6.5%).

Spectral measurements showed the lowest dispersion

for non-irradiated cells, while automated counting

was the best option for measuring the viability of

irradiated cells. Dispersion of viability measured by

cytometry was quite large with a median value close

to 10%. Considerations about dispersions will be

discussed in section 4.

The correlation between quality values and

viability, as measured by either the automated counter

or the cytometer was reported (figure 7). To avoid

cluttering the figure in the (100,100) coordinate

BIODEVICES 2025 - 18th International Conference on Biomedical Electronics and Devices

region, only data corresponding to irradiated cells

were showed in this figure.

Figure 7: Relation between quality and viability for (a)

automated counting and (b) cytometry.

Because of the dispersion shown in figure 6, the

data shown in figure 7 also showed a large dispersion.

The dispersion was 8.6% for the (quality/counter)

results and 10% for the (quality/cytometer) results.

This aspect will be discussed in section 4.

Nevertheless, linear regressions could be calculated

with acceptable R

around 0.9. The linear regressions

for both cases are given below.

𝑄𝑢𝑎𝑙𝑖𝑡𝑦0.86𝑉𝑖𝑎𝑏



15.3

)

𝑄𝑢𝑎𝑙𝑖𝑡𝑦1.08𝑉𝑖𝑎𝑏







1.55

)

The difference between these regressions will be

discussed in section 4.

3.3 Comparison of Viability and

Quality of Cells Stored at Low

Temperature

Figure 8 shows examples of spectra recorded at

different temperatures. At 37°C, the concentration

normally increases from 5.6×10

to 7.6×10

cell.mL

in 24 hours (generation time: 35 hours). The shape

of the spectra remains constant over time. At 25°C,

the concentration decreases slightly from 5.5×10

5.1×10

cell.mL

-1

and the shape of the spectra starts

changing after 21 hours. At 4°C, the concentration

decreases sharply from 5.5×10

to 3.6×10

cell.mL

-1

while the shape of the spectra changes much earlier.

Figure 8: Examples of spectra recorded at (a) 37°C, (b)

25°C and (c) 4°C.

Figure 9 shows the evolution of the corresponding

viability and quality. The evolution of these values is

consistent with the spectra shown in figure 8.

Figure 9: Viability and quality evolution corresponding to

spectra shown in figure 7. (a) Viability measured with

automated counter, (b) Quality.

White Light Spectroscopy for Mammalian Cell Viability/Quality Assessment: Towards an Online, Label-Free and Sampling-Less System to

Simplify Quality Control in CAR T-Cells Production

Finally, figure 10 shows the linear correlation

between viability and quality for all data recorded at

25°C and 4°C (37°C was omitted to avoid cluttering

the figure in the (100,100) coordinate region).

Figure 10: Relation between viability (counter) and quality

(spectroscopy).

The data dispersion is high for the reasons

mentioned above. The linear regression is shown

below.

𝑄𝑢𝑎𝑙𝑖𝑡𝑦1.13𝑉𝑖𝑎𝑏



3.75

(5)

In conclusion, according to the results shown in the

above sections, a relationship exists between quality

measured spectrally and viability measured by either

conventional methods. This relationship is linear and

depends on both the conventional method used as a

gold standard and the death process the cells undergo.

White light spectroscopy can therefore be used to

measure the viability of a cell suspension but should

apparently be adapted to experimental conditions.

However, the quality value may also be an alternative

method to estimate the cell health as discussed in

section 4.

4 DISCUSSION

Data Format

The original experimental setup was used to measure

cell concentrations. A discussion of data format was

proposed to explain why spectra were presented in

terms of absorbance rather than other formats such as

transmission or optical densities (OD) (Wacogne,

2022). In the current work, spectra were recorded in

the same way to facilitate mathematical treatments.

Note that the word "absorption" can sometimes be a

misnomer when considering small size particles, as

discussed later on.

Temperature-Induced Death Data

Experiments were performed in two ways: (i) cells

were prepared in the morning, stored at different

temperatures and transmission spectra were recorded

every 2 hours until the next morning, (ii) cells were

prepared in the evening, one transmission spectrum

was recorded in the evening and the others the next

day. Since viability decreases quite slowly when cell

death is induced by temperature stress, nothing

happens on the first day if the experiment was started

in the morning. Therefore, the examples shown in

figures 8 and 9 correspond to experiments started in

the evening.

Light-Matter Interaction, Spectra Shapes and

Remarks Concerning Optical Densities

The light-matter interaction process depends on the

size of the illuminated particles (figure 11).

Figure 11: Light-matter interaction with particles.

When the particles are large (around 10µm), the

light propagates mostly straight and there is very little

diffusion. In this case, the dominant light-matter

interaction process is absorption by cell constituents.

Spectrally, the shape of the absorption spectrum is

approximately Gaussian, at least for T-cells (figure

2(a)).

For smaller particles (around 1 µm), there is little

absorption and diffusion predominates. In this case,

diffusion evolves according to 1/λ, which explains

the spectra shape when large and small particles are

present (figure 2(b)).

For even smaller particles, diffusion is clearly

dominant and evolves according to 1/λ

Finally, for molecular size events, the medium can

be considered homogeneous and only absorption

occurs according to the molar absorptivity of the

molecules.

In this work, cell suspensions with high viability

correspond to the first row of figure 11, while

BIODEVICES 2025 - 18th International Conference on Biomedical Electronics and Devices

suspensions with low viability (regardless of the

death induction method) correspond to the second

row of figure 11. It is therefore inappropriate to speak

of absorption when the dominant light-matter

interaction is diffusion. In fact, the light not incident

on the spectrometer has not only been absorbed, but

mainly diffused. In this case, the word attenuation

may be preferred (Wacogne, 2023).

Also, the term optical density theoretically refers

only to the absorption of homogeneous media (fourth

row in figure 11). In other situations, such as

measuring the concentration of a bacterial

suspensions using a plate reader, the term optical

density is inappropriate. Nevertheless, the Beer-

Lambert law can still be applied (especially the law

of OD additivity for reference taking), but a term

other than OD should be preferred.

Data Dispersion

Dispersion is clearly observed in viability and quality

measurements. Dispersion can have several causes, as

already mentioned (Wacogne, 2022, 2023) for cell

concentration measurements. With conventional

techniques, viability measurements are performed

with very small volumes (a few tens of µL), making

the measurement poorly representative of what is

actually present in the culture cuvettes. The low-cost

plastic cuvettes used in this work have variations in

their optical properties that lead to inaccuracies of a

few percent.

The results presented here suggest that the

dispersion was even greater when viability was

measured. This is probably due to the difficulty for

conventional systems to actually determine whether a

cell is alive or dead. This results in a large dispersion

of data when the relationship between quality and

viability measured by either conventional mean is

reported (figures 7 and 10). This dispersion is not due

to possible inaccuracies in the measurement of quality

by spectroscopy since optical spectroscopy proved to

be more accurate than automated counter for cell

concentration measurements (Wacogne, 2021). We

could have then thought that quality measurements

would have shown less dispersion than other

conventional means. It is not the case. One hypothesis

is that equation (1) used to fit spectra could be

improved and/or that fitting intervals reported in table

1 could be adjusted. Indeed, the equation describing

the shape of the absorption spectra of CEM-C1 cells

includes a wavelength dependent base line not

accounted for in equation (1) (equation (5); Wacogne,

2022).

Relationships Between Quality and Viability

Figure 7 shows the relationships between quality and

viability when cell death is induced by irradiation for

2 gold standard techniques: automated counting and

cytometry. The relationships differ in terms of linear

regression slopes and ordinates at the origin. Figure

10 shows the relationship between viability, as

measured by automated counting, and quality when

cell death is thermally induced. It differs from the

counting/spectroscopy (irradiation) relationship, but

is similar to the cytometry/spectroscopy (irradiation)

relationship.

Viability measurement depends on several

factors. The discrepancy between methods is related

to either the predominant T-cell death process and/or

the type of biophysical interaction phenomena used

for viability measurement (Chung, 2015, Fotakis,

2006 and Chan, 2015). In fact, it depends on the type

of cell death and the ability of the viability

measurement equipment to detect it.

When cells die, they can undergo either apoptosis

or necrosis (autophagy is not discussed here). In its

early stages, apoptosis is characterised by a slight

reduction in cell size (a few µm). Later, the cells

disappear, producing micro- and sub-micrometre-

sized apoptotic bodies and microvesicles. In its early

stages, necrosis is characterised by a slight swelling

of cells. Later, cell membranes rupture and release

micro- and sub-micrometre-sized necrotic bodies.

Cell membrane debris remains in the suspension.

In this paper, viability is measured by cytometry

or automatic counting. The cytometric markers used

in this study makes it possible to distinguish apoptosis

from necrosis. Every not marked and cell-sized event

is considered as a living cell. The automatic counter

is based on trypan blue staining. This dye penetrates

cells but is expelled by living cells. It therefore stains

cells undergoing early necrosis. Early apoptotic and

late and large necrotic bodies are either stained or not

depending on individual death state. Late apoptotic

and small necrotic bodies are not detected.

The quality measured spectrally reflects the extent

to which the shape of the measured spectrum deviates

from a Gaussian shape. A decrease in this quality

value therefore reflects the presence of particles

smaller than healthy cells in the suspension, whatever

the type of cell death. It is mostly late apoptotic and

necrotic bodies, including cell membrane debris, that

modify the shape of the measured spectrum.

Regarding X ray induced cell death and to

understand the differences between the relationships

shown in figure 7, it is important to distinguish

between the ability of conventional viability

White Light Spectroscopy for Mammalian Cell Viability/Quality Assessment: Towards an Online, Label-Free and Sampling-Less System to

Simplify Quality Control in CAR T-Cells Production

measurement methods to actually produce accurate

results when viability is either high or low.

High viability range. The automated counter,

based on trypan blue staining, easily detects early

necrotic cells and hardly detects late necrotic bodies.

Some early apoptotic cells are detected while no late

apoptotic bodies are seen. This leads to an

overestimation of viability by automated counting

when viability is high, already reported previously

(Cai, 2023). To account for this, the data shown in

figure 7(a) should be slightly shifted to the left in the

high viability range, thus increasing the slope of the

relationship and decreasing the ordinate at the origin.

Low viability range. In this case, cytometry barely

detects dead cells. In fact, cells at this stage of death

are weakly expressing targets for AnV and 7AAD

markers, and both late apoptotic and necrotic bodies

are barely detectable. This leads to an overestimation

of viability by cytometry when viability is low.

Indeed, this is the reason why the apparent viability

increased between 24 and 48 hours post-irradiation

(figure 5(b)). To account for this, the data shown in

figure 7(b) should be shifted slightly to the left in the

low viability range, thus decreasing the slope of the

relationship and increasing the ordinate at the origin.

Taking all these corrections into account should

reduce and perhaps eliminate the differences in the

relationships shown in Figure 7 and equations (3) and

(4).

With regard to the effect of temperature, it is

difficult to interpret the relationship shown in figure

10 and equation (5) without comparison with other

measurement methods. The fact that the slope is equal

to 1 and the ordinate at the origin is equal to 0 does

not lead to any conclusion at this stage. It is not

possible to estimate the predominant T-cell death

process occurring at low temperature from these data.

However, it was reported that hypothermia mainly

induces apoptosis (Rauen, 1999; Wang, 2017).

Quality as a More General Alternative to Viability?

The measurement of viability remains open to

question since it depends on type of cell death and of

biophysical interaction phenomenon used in the

different measurement techniques. This explains the

differences observed in the literature when different

techniques are compared.

Apart from specific applications where the cell

death process is being studied, viability

measurements are mainly used to assess the health of

cell cultures, to evaluate experimental effects or to

optimise culture conditions. High cell viability

indicates a healthy and robust culture, whereas low

viability indicates problems with the culture

conditions or possible contamination. In these cases,

the spectrally measured quality value could provide a

more general alternative to viability measurement.

Indeed, the quality value indicates the extent to which

the cell suspension deviates from an ideal situation

where all cells are viable without relying to any bio-

physical-chemical interactions or equipment

specificity. Noted that early apoptotic and necrotic

bodies do not alter the spectra shape as much as late

bodies. An adjustment of equation (1) and/or table (1)

may be necessary to account for these slight spectral

changes.

Medico-Economic Considerations about CAR T-Cells

This study may have applications into the field of

CAR T-cell production and in particular the

expansion phase of several days, with each additional

day increasing their cost. Quality controls are

currently used to follow cell expansion but requires

frequent sampling, increasing risk of contamination.

The results outlined above are particularly

noteworthy because they enable quality control

without the need for sampling and allow for a rapid

termination of the expansion phase if issues arise.

CAR T therapy, designed to treat patients with

currently incurable diseases, has the potential to

revolutionize treatment in the coming years.

However, it remains still challenging to predict the

range of diseases these therapies may address or how

many patients might ultimately benefit from them.

Currently, there are several CAR T therapies on

the market (Wang, 2023; Bogert, 2021), each with

high price tags due to the complexity of treatment and

production. Some of the prominent FDA-approved

CAR T therapies include:

• Kymriah (tisagenlecleucel): Used to treat acute

lymphoblastic leukemia (ALL) and certain

types of lymphoma. It costs approximately

$475,000 per dose.

• Yescarta (axicabtagene ciloleucel): Primarily

used for large B-cell lymphoma, with a cost of

around $373,000 per dose.

• Tecartus (brexucabtagene autoleucel):

Approved for mantle cell lymphoma, priced

similarly to Yescarta at around $373,000.

• Breyanzi (lisocabtagene maraleucel): Treats

large B-cell lymphoma and has a list price of

about $410,000 per dose.

• Abecma (idecabtagene vicleucel): Approved for

multiple myeloma, costing around $419,500 per

dose.

These therapies, while promising in treating

various blood cancers, also incur additional costs

BIODEVICES 2025 - 18th International Conference on Biomedical Electronics and Devices

related to hospitalization and managing side effects

like cytokine release syndrome (CRS). This often

pushes the total cost per patient toward or beyond $1

million (Bogert, 2021; The ASCOP Post, 2018).

Automating CAR-T-cell production could

significantly reduce costs by lowering the need for

specialized labour and minimizing errors, thus

addressing the high price of treatments. It would also

improve scalability, enabling faster production and

wider accessibility. Standardizing the process could

enhance consistency, leading to better therapeutic

outcomes. Automation would also shorten the

turnaround time, allowing patients to start treatment

sooner, reducing need for costly interim therapies.

Finally, automating production could lower

healthcare burdens and make CAR T more accessible.

Online Integration Possibilities

White light spectroscopy can be easily integrated into

an online system (Wacogne, 2020). This is directly in

line with the advantages of automating the

manufacture of CAR T mentioned above. In addition,

the use of extremely compact light sources (Yujileds)

and spectrometers of fingernail size (Hamamatsu)

should facilitate and accelerate the integration of this

method.

Thus, having an online, real-time system for

monitoring cell expansion, quality of cells produced

and, potentially, real-time detection of any

contamination would represent a definite added value

for the development of these production systems.

5 CONCLUSIONS

In this paper, white light spectroscopy is used to

estimate the viability of T-cell line suspensions using

a “quality value” determined from the shape of the

absorption spectra of cell suspensions. A direct

relationship is observed between this quality value

and the viability measured by various conventional

methods. However, this relationship depends on the

conventional system used and the way in which cell

death is induced. These discrepancies in viability

measurements has already been mentioned in the

literature when comparing different conventional

methods.

Measuring the “quality value” could provide an

alternative to viability assessment less dependent on

equipment and type of cell death involved because it

is based solely on analysis of spectra shape and

independent from complex biophysical-chemical

interactions.

The measurement of a single absorption spectrum

of lymphocyte suspensions makes it possible to

monitor cell concentration and possibly detect any

contamination, but also to assess quality/viability of

cell cultures during the production of CAR T

therapies, leading to the possible online integration of

white light interferometry. This possibility of

automation would then be a step towards reducing the

price of these innovative therapies and making their

use more democratic.

ACKNOWLEDGEMENTS

This work was supported by the MiMedI project

(Grant N° DOS0060162/00) and the BioIMP project

(Grant N° BFC000802) funded by EU through the

European Regional Development Fund of the Region

Bourgogne Franche-Comté.

REFERENCES

Agilent: https://www.agilent.com/en/product/cell-analysis/

real-time-cell-analysis/rtca-analyzers/xcelligence-rtca-

esight-imaging-impedance-741228?gad_source=1&gc

lid=Cj0KCQjw3vO3BhCqARIsAEWblcCSop2Rma9n

WV2mnlpm1rzBEmhuFsxR3WFYtNWWCzbMyZ6v

GHjDzBcaAtSMEALw_wcB&gclsrc=aw.ds

Beckman: https://www.beckman.fr/cell-counters-and-anal

yzers/vi-cell-xr

Berridge, M. V., et al, 2016. Tetrazolium dyes as tools in

cell biology: New insights into their cellular reduction.

Biotechnology Annual Review, Vol. 11, pp.127-152

Bogert, R. 2021. Improving Outcomes and Mitigating Costs

Associated With CAR T-Cell Therapy. Supplement to

the American Journal of Managed Care, Vol. 27, pp.

S253-S261

Cai, Y., et al, 2024. Assessment and comparison of viability

assays for cellular products, Cytotherapy, Vol. 26, pp.

201-209

Chan, L.L., et al, 2015. Morphological observation and

analysis using automated image cytometry for the

comparison of trypan blue and fluorescence-based

viability detection method, Cytotechnology, Vol. 67,

pp. 461-473

Chung, D.M., et al, 2015. Evaluation of MTT and Trypan

Blue assays for radiation-induced cell viability test in

HepG2 cells. Int J Radiat Res, Vol. 13, pp. 331-335

Fotakis, G., et al, 2006. In vitro cytotoxicity assays:

Comparison of LDH, neutral red, MTT and protein

assay in hepatoma cell lines following exposure to

cadmium chloride, Toxicology Letters, Vol. 160, pp.

171-177

Fuente-Jiménez, J.L., et al, 2023. A comparative and

Critical Analysis for In Vitro Cytotoxic Evaluation of

Magneto-Crystalline Zinc Ferrite Nanoparticles Using

White Light Spectroscopy for Mammalian Cell Viability/Quality Assessment: Towards an Online, Label-Free and Sampling-Less System to

Simplify Quality Control in CAR T-Cells Production

MTT, Crystal Violet, LDH, and Apoptosis Assay. Int.

J. of Mol. Sci., Vol. 24, PP. 12860

Hamalainen-Laanaya, H.K., et al, 2012. Analysis of cell

viability using time-dependent increase in fluorescence

intensity. Analytical Biochemistry, Vol. 429, pp. 32-38

Hamamatsu: https://www.hamamatsu.com/eu/en/product/

optical-sensors/spectrometers/mini-spectrometer.html

Kumar, G., et al, 2015. Flow cytometry evaluation of in

vitro cellular necrosis and apoptosis induced by silver

nanoparticles, Food and Chemical Toxicity, Vol. 85,

pp. 45-51

Logos Biosystems: https://logosbio.com/

Louis, K.S., et al. 2011. Cell Viability Analysis Using

Trypan Blue: Manual and Automated Methods.

Stoddart, M. (eds) Mammalian Cell Viability. Methods

in Molecular Biology, Vol. 740, pp. 7-12

Macklin, C.C., et al, 1920. A study of brain repair in the

ratby the use of trypan blue: with special reference to

the vital staining of the macrophages. Archives of

Neurology and psychiatry, Vol. 7, pp. 353-NP

Ong, L.J.Y., et al, 2020. Quantitative Image-Based Cell

Viability (QuantICV) Assay for Microfluidic 3D Tissue

Culture Applications. Micromachines, Vol.11, pp. 669

Optiz, C., et al, 2019. Rapid determination of general cell

status, cell viability, and optimal harvest time in

eukaryotic cell cultures by impedance flow cytometry.

Applied Microbiology and Biotechnology, Vol. 103, pp.

8619-8629

Rauen, U., et al, 1999. Cold-induced apoptosis in cultured

hepatocytes and liver endothelial cells: mediation by

reactive oxygen species. Faseb journal, Vol. 13,

pp.155–168

Sarma, K.D., et al, 2000. Improved sensitivity of trypan

blue dye exclusion assay with Ni2+ or Co2+ salts.

Cytotechnology, Vol. 32, pp. 93–95

Schneckenburger, H., et al, 2012. Light exposure and cell

viability in fluorescence microscopy. Journal of

Microscopy, Vol. 245, pp. 311-318

Shenkin, M., et al, 2007. Accurate assessment of cell count

and viability with a flow cytometer. Cytometry Part B:

Clinical Cytometry, Vol.72B, pp. 427-432

Stoddart, M. J. 2011. Cell Viability Assays: Introduction.

Methods in Molecular Biology, Vol. 740, pp. 1-6

System C Bioprocess: https://www.system-c-bioprocess.

com/produit/fluidlab-r-300/

The ASCO Post: https://ascopost.com/issues/may-25-

2018/weighing-the-cost-and-value-of-car-t-cell-

therapy/

Thermofisher: https://www.thermofisher.com/order/cata

log/product/L3224

Tsaouis, K., et al, 2013. Time-dependent morphological

alterations and viability of cultured human trabecular

cells after exposure to Trypan blue. Clinical &

Experimental Ophthalmology, Vol. 41, pp. 484-490

Wacogne, B.; et al, 2020, Optical Spectroscopy for the

Quality Control of ATMP fabrication: A new method

to monitor cell expansion and to detect contaminations,

In Proceedings of the 13th International Joint

Conference on Biomedical Engineering Systems and

Technologies (BIOSTEC 2020), Springer Vol. 1, pp.

64-72

Wacogne, B. et al, 2021. White light spectroscopy for T-

cell culture growth monitoring: towards a real-time and

sampling free device for ATMPs production. Journal of

Translational Science, Vol. 7, pp. 1-10

Wacogne, B., et al, 2022. Absorption Spectra Description

for T-Cell Concentrations Determination and

Simultaneous Measurements of Species during Co-

Cultures, Sensors, Vol. 22, art. 9223

Wacogne, B., et al, 2023. Absorption/attenuation spectra

description of ESKAPEE bacteria: application to

seeder-free culture monitoring, mammalian T-cell and

bacteria mixture analysis and contamination

description, Sensors, Vol. 23, art. 4325

Wang, J., et al, 2017. The analysis of viability for

mammalian cells treated at different temperatures and

its application in cell shipment. PLoS ONE 12, Vol.4,

pp. e0176120

Wang J.Y,, et al, 2023. CAR-T-cell therapy: Where are we

now, and where are we heading?, Blood Sci., Vol. 5, pp.

237-248.

Wang, X., et al, 2016. Clinical manufacturing of CAR T-

cells: Foundation of a promising therapy, Mol. Ther.

Oncolytics, Vol. 3, pp. 16015

Yang, B., et al, 2023. Label-Free Sensing of Cell Viability

Using a Low-Cost Impedance Cytometry Device.

Micromachines, Vol. 14, pp. 407

YUJILEDS : https://store.yujiintl.com/products/yujileds-

hyperspectral-350nm-1000nm-0-4w-led-smd-2835

Zhong, J., et al, 2021. Multi-frequency single cell electrical

impedance measurement for label-free cell viability

analysis. Analyst, Vol. 146, pp. 1848-1858

Zou, Y., et al. 2013. Application of LDH-Release Assay to

Cellular-Level Evaluation of the Toxic Potential of

Harmful Algal Species, Bioscience, Biotechnology, and

Biochemistry, Vol. 77, pp. 345–352

BIODEVICES 2025 - 18th International Conference on Biomedical Electronics and Devices