Impact of Processing Delays and Cryopreservation on Sensitivity and Specificity of the Lymphocyte Transformation Test

Abstract

Introduction

In the lymphocyte transformation test (LTT), peripheral blood mononuclear cells (PBMCs) from patients with presumed drug allergy are cultured with the implicated drug. No study has systematically investigated how delayed processing or cryopreservation of PBMCs affects the sensitivity and specificity of the LTT to detect a drug sensitization. We evaluated the impact of delayed processing or cryopreservation of PBMCs on the sensitivity and specificity of the LTT focusing on interferon-y (IFN-γ) secretion as read-out. Additionally, we investigated the impact of the respective processing methods on non-stimulated reference and TT-stimulated PBMCs.

Methods

Blood from 30 patients and 30 gender-matched controls was split for paired processing: one aliquot was processed immediately as internal reference of the patient/control, the other subjected to either a 24-h delay (n = 15) or immediate processing followed by cryopreservation of the PBMCs (n = 15). Thereby, each patient and control in the respective cohort served as its own reference for the assessment of the methodical processing impact. IFN-γ concentrations in PBMC culture supernatants were measured by ELISA.

Results

Cryopreservation of PBMCs resulted in a statistically non-significant trend toward higher sensitivity (45% vs. 32%) and specificity (81% vs. 79%), for detecting drug sensitization compared to delayed processing. Overall, 80% of the LTT results using cryopreserved PBMCs were consistent with those obtained from the immediately processed reference group.

Conclusion

Due to the lack of statistical significance, no clear recommendation for one of the two processing methods can be made.

Introduction

The lymphocyte transformation test (LTT) is an in vitro method to detect drug sensitization in suspected drug allergies. In this test, peripheral blood mononuclear cells (PBMCs) from a patient with a presumed drug allergy are cultured with the suspected drug for up to 6 days. In sensitized individuals, drug-specific memory T cells recognize the drug, become activated, undergo transformation into blasts, and begin to proliferate [1, 2]. The traditional read-out method is the detection of radioactivity (counts per minute, cpm), resulting from the incorporation of radio-labeled 3H-thymidine in proliferating T cells. Drug-specific T-cell proliferation serves as an indicator for drug-specific sensitization. A stimulation index (SI) value greater than 2, or greater than 3 in the case of beta-lactams, is considered as a positive result. The SI is defined as the cpm ratio of the drug-exposed culture and the non-exposed culture [1, 2].

In recent years, alternative read-out methods and parameters, such as the detection of cytokine secretion via ELISA, ELISpot, or flow cytometry, have emerged [3, 4]. The sensitivity of the LTT ranges from 25% to 89%, depending on the severity of the allergic reaction and the read-out parameter used. The specificity lies between 63% and 100%. The phenotype (pathophysiology) of the reaction (e.g., immediate [type I] or delayed [type IV]) and the drug being investigated also influence sensitivity and specificity [1, 5]. Due to these limitations and its complexity, the LTT is currently not part of the routine diagnostics and is only performed in laboratories with the necessary expertise and technical infrastructure. As a result, patient and control blood samples are often collected at the clinical site and then transported to the laboratory. Hence, delayed processing of the blood samples may exert detrimental effects on its performance in terms of sensitivity and specificity.

With regard to transport and sample processing, two principle approaches can be envisaged: (i) conduction of the LTT with one day of delay between blood collection and LTT setup due to blood sample transport and (ii) isolation and freezing of PBMCs at the clinical site, shipping the PBMCs to the LTT-performing laboratory, where they are then thawed and the LTT is performed. Multiple studies have investigated the effects of cryopreservation on PBMCs viability and functionality including the responsiveness to tetanus toxoid (TT), a strong antigen, which is used as a positive control for antigen-specific responses in the LTT [6]. However, drugs give a much weaker cellular response in the LTT as TT and the effects of different processing methods on the performance of LTT for the in vitro determination of drug sensitization have not systematically been investigated. An older review states that freshly processed blood showed similar results as blood stored for 24 h, but no data were shown [2]. A more recent study postulated that a delay in PBMCs isolation from blood may significantly decrease antigen-specific T-cell responses [7].

The aim of this real-life study was thus to investigate to what extent (i) a simulated longer transport (24 h) of blood samples and (ii) cryopreservation of the PBMCs immediately after isolation affects the performance of the LTT. Therefore, the primary parameters considered were the sensitivity and specificity of the LTT for the detection of drug sensitization. Additionally, the impact on IFN-γ secretion of unstimulated and TT-stimulated PBMCs was also assessed as these serve as background and positive control in the LTT, respectively.

Methods

Patients and Controls

Thirty patients (mean age 50 years [median: 53 years]; 19 female, 11 male) with reported drug allergies and 30 gender-matched control individuals without allergy to the tested drug were recruited from the Department of Dermatology and Allergy, University Hospital Aachen or the Federal Institute for Drugs and Medical Devices, respectively (Table 1). Drug sensitization was confirmed by a comprehensive medical history and at least one positive allergy test, either in vitro (such as a basophil activation test or drug-specific IgE) or in vivo (including skin tests or provocation tests) except for 3 patients recruited via BfArM (stringent medical history only). Although not all patients underwent the same tests, all tests required to reliably diagnose the drug allergy to the suspected drug were performed in all patients except for three patients in whom an unequivocal medical history alone confirmed the allergy.

Table 1.

Characteristics of patients and control individuals

Patient #	Sex	Age	Drug	Reaction phenotype (immediate (type I) or delayed (type IV))	Time span between reaction and blood donation	Sample handling	Control #	Sex	Age
1	f	61	Cefuroxime	I	6 months	24-h storage	1	f	62
			Ceftriaxone		n.a.
2	m	58	Amoxicillin	IV	15 years	Freezing	2	m	41
3	f	64	Clindamycin	IV	4 months	24-h storage	3	f	34
4	m	62	Cefuroxime	I	5 months	Freezing	4	m	41
5	f	68	Amoxicillin	IV	12 months	24-h storage	5	f	19
6	m	57	Cefuroxime	I	2 years	Freezing	6	m	43
7	m	23	Cefuxoime	I	7 months	Freezing	7	m	36
			Metamizole		6 years
8	f	62	Cefuroxime	I	12 months	24-h storage	8	f	61
9	f	52	Metamizole	I	19 months	24-h storage	9	f	38
10	m	64	Penicillin G	I	30 years	24-h storage	10	m	36
11	m	28	Metamizole	I	8 months	Freezing	11	m	37
12	f	55	Ampicillin	IV	1 month	Freezing	12	f	26
13	f	19	Metamizole	I	6 months	24-h storage	13	f	30
14	f	31	Clindamycin	IV	18 months	24-h storage	14	f	55
15	f	31	Ampicillin	IV	6 months	Freezing	15	f	29
			Amoxicillin
16	m	38	Cefuroxime	I	9 months	Freezing	16	m	42
			Amoxicillin
17	f	62	Cefuroxime	I	11 months	Freezing	17	f	59
18	f	43	Metamizole	I	6 months	24-h storage	18	f	24
19	f	33	Amoxicillin	IV	24 months	24-h storage	19	f	28
20	m	46	Amoxicillin	I	9 months	24-h storage	20	m	25
21	m	58	Clindamycin	IV	5 months	24-h storage	21	m	36
			Amoxicillin		24 months
22	m	59	Penicillin G	IV	>30 years	Freezing	22	m	37
23	f	43	Amoxicillin	IV	>20 years	Freezing	23	f	31
			Penicillin G
24	f	44	Amoxicillin	I	6 months	24-h storage	24	f	28
			Penicillin G
25	m	47	Amoxicillin	I	5 months	24-h storage	25	m	50
			Penicillin G
26	f	51	Cefuroxime	I	2 years	Freezing	26	f	28
27	f	38	Metamizole	I	5 months	24-h storage	27	f	28
28	f	65	Ampicillin	IV	>15 years	Freezing	28	f	27
			Amoxicillin
29	f	67	Cefuroxime	I	9 months	Freezing	29	f	27
30	f	66	Amoxicillin	IV	11 months	Freezing	30	f	27

All patients were evaluated while they were free of any clinical symptoms related to drug allergies. They were only included if they had not experienced any infections within a month prior to blood donation. The results of the patients for the immediately processed blood samples are partly mentioned in published studies of our group, which served a different objective than the present study [8]. For each LTT with a sensitized individual, an LTT with a sex-matched control was performed simultaneously.

Sample Processing

After blood collection (approximately 50 mL) into lithium heparin vacutainers (BD Vacutainer^®), the blood samples were immediately shipped to the laboratory site in less than 3 h. Upon arrival, blood samples were treated differently (Fig. 1).

Fig. 1.

Flowchart of the blood sample processing and analysis. The samples prepared with a 24-h delay (n = 15) are depicted on the left (cohort A) and samples where PBMCs were cryopreserved (n = 15) are depicted on the right (cohort B). Each cohort was compared to a respective reference group (references A and B), consisting of PBMCs from the same patients and controls, which were immediately processed. Hence, each patient and each control served as its own control for the assessment of the impact of the delayed processing and cryopreservation.

24-h storage (cohort A): Blood samples were split into two fractions. With the first fraction, PBMCs isolation and LTTs were performed immediately as described previously [9]. The second fraction was stored in the dark for 24 h in a locker at ambient temperature to simulate overnight delivery conditions. On the next day, the stored blood samples were treated identically to the immediately processed samples (section 2.3).

Freezing (cohort B): after PBMCs isolation, washing, and counting, at least 2.5∙× 10⁷ cells were separated, spun down (10 min, 400 rcf, 23°C), and resuspended in 1 mL RPMI 1610 medium supplemented with 12.5% autologous plasma and 10% DMSO [10]. The cell vials were then placed into an insulated cryo-freezing container, placed in a −80°C freezer, and cooled at a rate of approximately 1°C/min. Additionally, 10 mL of heat-inactivated and centrifuged autologous plasma were frozen and stored at −80°C until thawed for use in the LTT with the cryo-conserved PBMCs. On the following day, frozen cells were shifted to a liquid nitrogen dewar for long-term storage (for >2 weeks and <2 months). After storage, cells were thawed at 37°C for 5 min and diluted in 9 mL RPMI 1640 medium containing 5% autologous plasma. Cells were washed once (400 rcf, 10 min, 23°C), resuspended in 10 mL RPMI 1640 medium (5% autologous plasma), and counted. Subsequently, cells were diluted and the LTT was performed as described below (see section 2.3). Notably, each patient and each control served as its own control for the assessment of the impact of the delayed processing and cryopreservation.

Lymphocyte Transformation Test

A detailed description of the LTT can be found elsewhere [1–3, 5, 9]. In brief, PBMCs were extracted from blood samples by density gradient centrifugation using SepMate^™ – 50 tubes (STEMCELL^™ Technologies) and Lymphoprep^™ density gradient medium (STEMCELL™ Technologies), washed repeatedly with PBS, and resuspended in RPMI 1640 medium supplemented with 5% heat-inactivated autologous plasma. Subsequently, 2 × 10⁵ PBMCs were seeded into 96-well round-bottom plates and incubated with the suspected drug. Online supplementary eTable 1 (for all online suppl. material, see https://doi.org/10.1159/000551013) provides the list of drugs and their concentrations. For each experimental condition (unstimulated, TT stimulated at 0.1 µg/mL as positive control, drug stimulated), a total of six replicate wells were used. After an incubation period of 6 days, PBMCs viability was assessed using trypan blue (viability >95% for all samples) [11], and the supernatants of the replicates were pooled and stored at −80°C until ELISAs were conducted.

ELISAs

ELISAs were performed using the ELISA MAX Deluxe human IFN-γ kit (BioLegend^®) or IL-5 kit (BioLegend^®). As described elsewhere [12], the detection limit for the IFN-γ ELISA was determined experimentally to be 1 pg/mL. The sensitivity limit was verified by extending the manufacturer’s reference standard range by additional concentrations of 3.9, 2, and 1 pg/mL. To account for the detection limit of 1 pg/mL, samples with an IFN-γ or IL-5 concentration below 1 pg/mL were set to 1 pg/mL. An ELISA was considered positive, if the SI, which is defined as the IFN-γ/IL-5 concentration of the drug stimulated sample divided by the IFN-γ/IL-5 concentration of the corresponding unstimulated sample, was ≥3.0 for beta-lactam antibiotics and ≥2.0 for non-beta-lactam antibiotic drugs [1, 2]. The culture with the drug concentration yielding the highest IFN-γ or IL-5 concentration, respectively, was used for the SI calculation. Notably, and in concordance with previous studies [9, 12], IFN-γ secretion levels were generally higher than for IL-5, where only few patients and controls in both study cohorts A and B showed a detectable change in the IL-5 secretion upon drug stimulation. Therefore, all subsequent analyses focus on the IFN-γ concentrations.

Statistical Analyses

Statistical analyses were conducted using the open-source software R (version 4.4.0, packages: tidyverse, ggpubr, boot). Data of TT-stimulated samples were analyzed using non-parametric paired comparisons between conditions (Wilcoxon signed-rank test, two sided).

For the intra-cohort comparison of the drug-stimulated samples, the corrected McNemar test was used. For the inter-cohort comparison (delayed vs. cryopreserved), Fisher’s test was applied and for the power analysis power.prop.test in R was used.

Results

Cell Counts and Viability

The average number of cells after PBMCs isolation was 1.6 × 10⁶ and 2.0 × 10⁶ per mL blood in the immediately processed (n = 30) and delayed processed blood samples (n = 15), respectively. After thawing of cryopreserved PBMCs (n = 15), the average recovery rate was 69% (Table 2). Representative photos of the trypan blue staining of immediately processed reference cells compared to the delayed-processed and cryopreserved cells after 144 h incubation in the LTT are shown in online supplementary eFigure 1. For each sample processing method, the cell viability of PBMCs after 144 h incubation in the LTT was >95% in the case of immediately and delayed processed blood and >90% for cryopreserved PBMCs, irrespective of the drug used for stimulation (Table 2).

Table 2.

Mean cell counts, IFN-γ concentrations, and SIs depending on the processing method

​	Transport and storage of blood sample, isolation of PBMCs
	blood sample (n = 30)	Cohort A: blood sample (n = 15)	Cohort B: blood sample (n = 15)
Transport to lab	<4 h	<4 h	<4 h
Storage	No	At room temperature 24 h	Freezing and cryopreservation of PBMCs for >2 weeks and <2 months
PBMC isolation	In the laboratory (<2 h) after receipt of blood sample	Ca. 24 h after blood sampling	In the laboratory (<2 h) after receipt of blood sample, then frozen
LTT	Immediately after PBMC isolation	Immediately after PBMC isolation	Immediately after thawing of PBMs
Cells/mL blood (mean, SD)	1.6 × 106 (±0.6 × 106)	2.0 × 106 (±0.7 × 106)	Not applicable1; recovery = 69±10%
Cell viability	>95%	>95%	>90%
IFN-γ concentration (mean, SD) of unstimulated samples, pg/mL	Cohort A: 31 (SD = 77)	16 (SD = 36)	25 (SD = 37)
	Cohort B: 21 (SD = 50)
IFN-γ concentration (mean, SD) of TT-stimulated samples, pg/mL	Cohort A: 4,318 (SD = 3,468)	3,622 (SD = 3,171)	3,086 (SD = 3,333)
	Cohort B: 3,809 (SD = 3,816)
SI of IFN-y secretion (mean, SD) to TT	Cohort A: 1,390 (SD = 1,987)	1,094 (SD = 1,728)	693 (SD = 1,570)
	Cohort B: 821

¹Not applicable as these PBMCs are simultaneously isolated with the immediately processed PBMCs. Instead, the recovery rate of viable PBMCs was chosen to assess the method performance.

Notably, for the statistical analysis of the IFN-γ concentrations and SI values of the frozen PBMCs, both controls #29 and #30 were renounced and not included in further analysis as they were identified as outliers due to increased IFN-γ concentrations of the unstimulated samples (online suppl. eTable 2).

IFN-γ Secretion of Unstimulated and TT-Stimulated PBMCs

Following TT stimulation, for all processing methods an increased IFN-γ concentration compared to the respective unstimulated PBMCs and therefore an increase of the SI >2 in all samples was observed. The effect of the different processing methods can be assessed (i) in terms of the absolute IFN-y concentration and (ii) in terms of the SI.

The IFN-y concentration of the unstimulated samples was generally low (Table 2) and no significant reduction or increase of the IFN-y concentration in the individual samples depending on the processing method was observed (online suppl. eFig. 2). In contrast, the effect of the processing method was more pronounced for the TT-stimulated samples (Fig. 2).

Fig. 2.

Changes of IFN-γ concentration following stimulation with TT depending on sample processing method. a Paired line plots showing the IFN-γ concentration change of the immediately processed sample (left) and delayed processed (right) samples for a 24-h delayed processing (cohort A, left, orange) or cryopreservation of the PBMCs (cohort B, right, blue) after TT stimulation. For clarification, decreasing concentrations are shown as light and downward pointing, increasing concentrations as dark and upward pointing triangles. b Violin plots showing the difference in IFN-γ concentration Δc(IFN-γ) between the TT-stimulated delayed processed or the cryopreserved samples and their respective reference samples using the same color coding as in a. Mean concentration differences and the confidence interval are shown as black square and gray boundaries. b shows a statistically significant reduction (Wilcoxon p = 0.000743) of the mean concentration difference Δc(IFN-γ) of the TT-stimulated samples after a 24-h delayed processing versus the immediately processed samples. This difference was not significant for the cryopreserved samples.

Delayed processing of the blood led to a reduced IFN-γ secretion in 25 of 30 TT-stimulated samples compared to the immediately processed reference sample. Cryopreservation affected the IFN-γ secretion to a lesser extent and showed no clear trend with 16 out of 28 samples showing a reduced IFN-γ concentration after TT stimulation (Fig. 2a). Consequentially, the mean concentration difference Δc(IFN-γ) of the TT-stimulated samples after a 24 h delayed processing versus the immediately processed samples showed a statistically significant reduction (Wilcoxon p = 0.000743). In contrast, this difference was not significant for the cryopreserved samples (Wilcoxon p = 1, Fig. 2b).

Notably, these changes in the IFN-γ concentration depending on the processing method do not directly translate into a statistically significant change of the mean SI for TT-stimulated samples as both, the 24 h storage as well as the cryopreservation of PBMCs, yielded similar fluctuations of the calculated SI values (60% and 46% decreased SI values for the delayed processed and cryopreserved samples, respectively; cohort A: p = 0.233, cohort B: p = 0.531, online suppl. eFig. 3). The reduced impact of the IFN-γ concentration change after TT treatment on the SI of the delayed processed samples can be rationalized by the subtle changes in the IFN-γ concentration of the unstimulated background samples on an individual level. Here, a smaller absolute change can still result in an increased SI as the ratio instead of the absolute values of the stimulated versus the unstimulated IFN-γ concentrations are used for the SI calculation.

IFN-y Secretion of PBMC after Incubation with the Suspected Drug

Next, we analyzed the LTT performance for both patient cohorts with respect to the IFN-γ secretion, the SI, and the consistency of the results using drugs as antigens. Since there are different possibilities to define positivity in the LTT with IFN-y read out, we also calculated the difference of the concentration of the drug-stimulated versus the unstimulated cultures (Δc(IFN-γ)_drug-unstim) for the immediate, delayed, and cryopreserved cultures (online suppl. eTable 3). However, the results did not provide a substantially different picture compared to the absolute IFN-y concentrations (Table 3).

Table 3.

Summary of the LTT results for delayed processing (left) and cryopreservation (right)

24 h storage (cohort A)				Cryopreservation (cohort B)
patient #	IFN-γ, pg/mL		trend	patient #	IFN-γ, pg/mL		trend
	immediate	delayed			immediate	delayed
8	26.5	28.6	↑	2	11.2	24.5	↑
18	7.6	67.2	↑	7d	9.9	21.6	↑
20	80.8	203.9	↑	7f	6.0	14.1	↑
21a	151.6	209.2	↑	12	82.9	88.2	↑
25a	36.1	51.1	↑	15g	184.0	409.6	↑
1b	1.0	1.0	–	15a	219.1	389.2	↑
9	2.1	2.1	–	16d	1.3	17.4	↑
10	1.6	1.6	–	16a	1.4	3.4	↑
24c	1.0	1.0	–	17	28.6	37.0	↑
1d	3.5	1.4	↓	22	9.5	21.9	↑
3	2.3	1.0	↓	26	2.1	20.5	↑
5	3.2	2.4	↓	29	5.2	55.1	↑
13	34.4	2.7	↓	30	7.3	42.2	↑
14	6.1	2.6	↓	4	5.5	1.9	↓
19	26.9	11.9	↓	6	17.4	8.9	↓
21e	18.3	9.4	↓	11	20.2	15.4	↓
24a	2.1	1.0	↓	23c	3.4	1.7	↓
25c	21.1	6.2	↓	23a	1.9	1.0	↓
27	1,328	788	↓	28g	215.1	10.5	↓
​	​	​	​	28a	26.5	12.6	↓

The numbers in the colored panels represent the highest IFN-γ concentration observed for the respective drugs for each patient. Light green and light red colors indicate one SI-positive and one negative LTT while dark green and dark red colors indicate both LTTs (immediate and delayed/cryopreserved) to be positive or negative, respectively. The arrows in the very right panel of a section highlight an increase (arrow up), decrease (arrow down), or no change (dash) of the IFN-γ concentration following the delayed/cryopreserved processing with respect to the immediate reference sample.

^aAmoxicillin.

^bCeftriaxone.

^cPenicillin G.

^dCefuroxime.

^eClindamycin.

^fMetamizole.

^gAmpicillin.

Of the 19 immediately processed reference samples in cohort A, 9 LTTs showed a positive SI (47% sensitivity, 63% specificity). For the immediately processed reference samples of cohort B (n = 20), also a total of 9 LTTs were SI positive (45% sensitivity, 65% specificity). As the sensitivity and specificity were comparable with only slight variations for a positive LTT, both cohorts were considered representative for the subsequent comparison of a delayed versus cryopreserved processing of the blood or PBMCs, respectively.

In cohort A, the sensitivity of the LTT declined to 32% (6/19 positive) while a gain in specificity to 79% was observed (15/19 controls negative, online suppl. eTable 4). Both the decrease in sensitivity and the simultaneous increase in specificity can be rationalized by a trend toward reduced total IFN-γ concentrations for delayed processing of untreated PBMCs where in 14 out of 19 LTTs the IFN-γ concentration was either reduced or unchanged compared to the results of the immediately processed blood samples. In other words, only LTTs with strong IFN-γ secretion remained positive, while weakly positive LTTs – both in patients and controls – were excluded, thereby reducing sensitivity but increasing specificity. This is in accordance with our findings using the strong antigen TT, where delayed processing led to a similar effect, albeit here it is less pronounced due to a generally lower IFN-γ secretion of the drug-stimulated samples compared to the TT-stimulated positive controls. Hence, SIs were generally lower following delayed processing, and the sensitivity was also reduced. In contrast, cryopreservation of PBMCs of cohort B did not affect the sensitivity of the LTT, with the same sensitivity of 45% (9/20 positive) as the immediately processed reference samples observed. In addition, specificity (81%) increased. Accordingly, for 13/19 samples, the IFN-γ concentration of the cryopreserved samples increased compared to the immediately processed reference samples. A detailed summary of the obtained sensitivities and specificities in dependence of the processing method and phenotype is given in online supplementary eTable 5.

Both processing methods showed largely consistent results compared to their reference samples with respect to the outcome of the LTT in terms of the SI. For the delayed processed samples, 74% of LTTs showed the same result compared to the immediately processed reference. For the cryopreserved PBMCs, 80% of the LTTs were consistent with the result of the LTTs of the immediately processed reference. However, for the remaining 26% LTTs (5 in total) of the delayed cohort A, in 4 of the 5 cases the delayed processed samples showed a negative LTT result and only one LTT became positive after a delayed blood processing. In contrast, of the 20% LTTs (4 in total) with cryopreserved samples showing a different outcome, 2 of 4 turned negative while the remaining two LTTs became positive despite a negative result of the immediately processed reference. However, the differences for both processing methods compared to immediate processing with regard to sensitivity were not significant using the continuity-corrected McNemar test (p: cohort A: 0.3711; cohort B: 1) and specificity (p: cohort A: 0.4795; cohort B: 0.3711).

Although freezing and thawing of the PBMCs shows a trend toward a more reproducible LTT outcome compared to delayed processing of the blood, this difference was not statistically significant (Fisher’s exact test, p = 0.5145; R = 0.57 [0.12–2.49]). A power-analysis (power 80%, applying the sensitivities obtained in both cohorts) revealed that the patient pool would need to be expanded to >200 patients in both cohorts to allow for the detection of a statistically significant result (online suppl. eAppendix).

Discussion

This is the first study to systematically investigate the impact of delayed blood sample processing and PMBC cryopreservation on the performance of the LTT used for the detection of drug sensitization. Cryopreservation of PBMCs resulted in a statistically nonsignificant trend toward higher sensitivity and specificity for detecting drug sensitization compared to delayed processing. However, the lack of statistical significance does not permit a clear recommendation for cryopreservation. Notably, cryopreservation cannot improve the overall performance of the LTT.

Effects of Both Processing Methods in TT-Stimulated Samples

The consistently low IFN-γ concentrations in unstimulated samples likely explain the absence of processing-related changes in this condition (Table 2, online suppl. eFig. 2). In contrast, delayed processing significantly reduced IFN-γ secretion upon TT stimulation, whereas cryopreservation did not. These findings suggest that delayed processing negatively affects antigen-specific T-cell responses, while cryopreservation appears more robust in preserving LTT reactivity (Fig. 2b).

This finding is in accordance with literature reporting that cryopreservation may slightly lower the SI but does not interfere significantly [10]. Likewise, although the mean proliferative response was decreased in frozen PBMCs stimulated with TT compared to TT-stimulated immediately processed PBMCs, the decrease was not statistically significant [13]. Notably, these changes in the IFN-γ concentration do not translate into a statistically significant change of the SI although 60% of the SI values decreased for the delayed processed and 46% for the cryopreserved samples.

Effects of Both Processing Methods in Drug-Stimulated Samples

In our analysis, cryopreservation resulted in a statistically nonsignificant trend toward higher sensitivity (45% vs. 31%) and specificity (81% vs. 79%) for detecting drug sensitization compared to delayed processing. Likewise, 80% of the LTT with cryopreserved PBMCs were consistent with the immediate-reference group in terms of LTT outcome. However, to confirm a statistically significant difference, the patient cohorts would need to be largely expanded (>200 in both cohorts). These figures are far beyond the typical numbers of patients being feasible to be included in an LTT study.

We cannot compare our results with other respective studies since this is the first study systematically investigating the impact of processing methods on the sensitivity and specificity of the LTT in terms of the SI defining positivity. Our results suggest that the consistency of a positive or negative LTT is not substantially changed as 74% of the delayed processed samples and 80% of cryopreserved PBMCs showed the same result as the immediately processed reference. However, we observed increased IFN-γ concentration in 13/19 of the cryopreserved samples compared to the immediately processed reference samples. Likewise, we noticed an increased difference of the IFN-y concentration of drug-stimulated minus the unstimulated cultures (Δc(IFN-γ)_drug-unstim) in 12/19 cryopreserved samples (online suppl. eTable 3).

The definition of a positive LTT with IFN-y secretion as a read-out has not been standardized and lacks consistency [12]. Different approaches have been published including, e.g., the calculation of a SI or focusing on the absolute increase in cytokines [12, 14]. While an SI is generally considered robust, it may lead to misclassification if the unstimulated sample has a moderate to high basal IFN-γ secretion. In fact, the basal IFN-y secretion can vary (online suppl. eTable 3). Hence, the selection of the method defining positivity in the LTT is of some importance when dealing with delayed or cryo-conserved samples.

Sensitivity of the LTT

Cryopreservation and delayed processing are intended to address a logistic problem if the samples cannot be processed at the site where the blood is taken. Notably, the aim of our study was not to evaluate whether both methods improve the sensitivity or specificity of the LTT and thereby improve its value for detecting drug sensitization. Instead, the aim was to show whether sensitivity or specificity of the LTT is affected by either pre-processing method. Our results show an identical sensitivity of 45% for identification of patients with drug sensitization in cohort B with cryopreserved PBMC and immediately processed blood samples (see online suppl. eAppendix). Hence, while cryopreservation of PBMCs may mitigate or neutralize detrimental effects, it does not address the inherent limitations of the LTT method itself. Consequently, cryopreservation cannot improve the overall performance of the LTT. Likewise, delayed processing cannot improve the sensitivity of the LTT and resulted in an even lesser sensitivity (31.6%) compared to immediately processed blood samples (47.4%) (cohort A, see online suppl. eAppendix).

The sensitivities and specificities of the LTT seen in our study are at the lower and middle range, respectively, of what is published in literature (45% versus 25–85% and 81% versus 63–100%, respectively) [1, 5]. Therefore, our data further corroborate the literature, suggesting that, due to its limitations in performance and inherent complexity, the LTT in its current form is not integrated into routine diagnostics.

Limitations and Strengths

Our study results are based on 15 LTTs with delayed processing and 15 LTTs with cryopreserved PBMCs, compared to the immediately processed control sample. Although a higher number of LTTs would be desirable from a statistical point of view, the number of n = 15 LTTs is relevant as LTTs require a significant amount of hand-on work and are not performed in routine laboratories. For the herein tested patient pool, the sensitivity (46%) was in the lower range of values reported in literature probably reflecting the higher proportion (18/30) of patients with immediate-type reactions, for which the LTT is less sensitive [5].

Notably, besides IFN-y other cytokines, like IL-2, -4, -5, -13, granulysin, or granzyme B, often two or three of them combined, are used for the LTT with cytokine read-out [4, 14–17]. However, it appears that IFN-y is the cytokine most frequently used, which is why we selected it for our study. We also investigated IL-5 in all LTTs. However, since the LTTs with IL-5 were rarely positive (see online suppl. eTable 6), we focused on IFN-y.

We did not quantify IFN-γ and IL-5 concentration baseline levels in the blood samples of each patient and control. For those few tested (n = 2 patient/control pairs in this study and n = 6 patient/control pairs in another study [not published yet]), the concentration levels for both cytokines were always well below the concentration in the unstimulated samples (data not shown). Additionally, increased IFN-γ and IL-5 levels prior to stimulation should not affect the outcome of the direct comparison of the handling methods as the same amounts would be present at the start in both, the immediately and the delayed processed or cryopreserved samples. Finally, our results pertain to the cytokine read-out of the LTT with IFN-y and IL-5. We cannot foresee whether our results can be extrapolated to the proliferation-based read-out of the LTT using 3H-thymidine incorporation or the cytokine-based read-out of the LTT using other cytokines.

Although not all patients underwent all tests to diagnose a drug allergy to the suspected drug, the necessary tests required to reliably diagnose the drug allergy were performed in all patients except for 3 patients in whom an unequivocal medical history alone confirmed the allergy. Hence, we do not see variability between the patients in terms of the reliability of the diagnosis with drug allergy being more likely in 1 patient compared to another. In addition, since there was no preferential assignment to a specific group (e.g., all patients with provocation tests exclusively in group A), we cannot see a selection bias toward one of the groups. The strength of our investigation is the fact that each patient served as its own individual control, and we evaluated performance on a patient and not averaged level.

Implications for Clinical and Laboratory Practice

Cryopreservation of PBMCs resulted in a statistically nonsignificant trend toward higher sensitivity (45% vs. 32%) and specificity (81% vs. 79%) for detecting drug sensitization compared to delayed processing. This trend seen with drug-stimulated PBMC is to some extent supported by the results seen with TT-stimulated PBMC where delayed processing significantly reduced IFN-γ secretion upon TT stimulation, whereas cryopreservation did not (Fig. 2b). Likewise, a recent study aiming at a different purpose, i.e., the modification of the LTT, postulated in the discussion of their results that a delay in PBMCs isolation from blood may significantly decrease antigen-specific T-cell responses [7]. However, since our study is the first systematically investigating the impact of processing methods on the sensitivity and specificity of the LTT for detecting drug sensitization we cannot put our findings in context with other findings, which may point in a similar or different direction. Hence, as a concluding statement, based on our results, considering the lack of statistical significance a clear recommendation on which method should be preferred in clinical practice cannot be given.

Acknowledgments

We thank both patient and control individuals for their donation of blood. During the preparation of this work, the English wording of the manuscript was polished by the large language model ChatGPT version: GPT-4.1 mini (OpenAI, San Francisco, CA, USA) and initial R scripts were derived from ChatGPT. The authors reviewed and edited the content and take full responsibility for the final text and R scripts.

Statement of Ethics

The study was approved by the Ethics Committee of the university hospital Aachen (study No. EK 309/19 and EK 23-187) and the medical council North Rhine (study Nos. 2020098 and 2023231). Written informed consent was obtained from the participants for participation in the study and publication of the details of their medical details.

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

The manuscript was written in context with an experimental study related to the improvement of the lymphocyte transformation test, which was funded by the ZukunftBIO.NRW program (funding No.: ZM-1-24B). Additional funding was obtained by own resources of the Federal Institute for Drugs and Medical Devices (BfArM), the Department of Dermatology and Allergy, University Hospital Aachen, and the Life & Brain Center, Department of Genomics, University of Bonn, Bonn, Germany. The funder had no role in the design, data collection, data analysis, and reporting of this study.

Author Contributions

Conceptualization: A.G., B.S., and C.H. Funding acquisition: P.H., A.Y., M.N., and B.S. Resources: A.Y., G.W., P.H., and M.N. Conception of experiments: A.G., C.H., and B.S. Investigation: C.H., A.G., D.D., and L.G. Visualization: C.H. Statistical data analyses: C.H., D.D., L.G., P.D., and B.S. Project Administration: P.D. Supervision: A.Y., P.H., M.N., B.S., and G.W. Validation and writing – review and editing: all authors. Writing – original draft: B.S. and C.H. All authors have approved the final version of the manuscript to be published and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Funding Statement

The manuscript was written in context with an experimental study related to the improvement of the lymphocyte transformation test, which was funded by the ZukunftBIO.NRW program (funding No.: ZM-1-24B). Additional funding was obtained by own resources of the Federal Institute for Drugs and Medical Devices (BfArM), the Department of Dermatology and Allergy, University Hospital Aachen, and the Life & Brain Center, Department of Genomics, University of Bonn, Bonn, Germany. The funder had no role in the design, data collection, data analysis, and reporting of this study.

Data Availability Statement

The data that support the findings of this study are not publicly available due to privacy reasons but are available from the corresponding author upon request.