Diagnostic performance of lymphocyte transformation test and flow cytometry in the identification of culprit drugs in organ-specific serious adverse drug reactions

What is already known about this subject

1. In vivo diagnostic methods such as skin tests and drug provocation tests are often unsafe or contraindicated in serious adverse drug reactions (SADRs).

2. Lymphocyte transformation test (LTT) and flow cytometry have been used independently to assess drug-specific T-cell activation.

3. Diagnostic sensitivity of individual in vitro assays is limited and highly variable depending on drug and reaction type.

What this study adds

1. The combined use of LTT and CD69 flow cytometry significantly increases diagnostic sensitivity across a wide range of organ-specific SADRs.

2. Diagnostic performance improves when in vitro results are interpreted within the framework of validated causality scoring systems.

3. These findings support the use of multimodal in vitro approaches as part of a comprehensive and safe diagnostic strategy in clinical pharmacology.

Study design and setting

This prospective observational study was conducted at La Paz University Hospital (Madrid, Spain), a tertiary referral centre. At this centre, a prospective Pharmacovigilance Program from Laboratory Signals in Hospital (PPLSH) has been implemented to proactively detect SADRs [12]. This program also managed consultations for suspected serious adverse events. The study period extended from January 2022 to July 2024. The research protocol was approved by the Institutional Ethics Committee (Code PI-3226; 25 May 2018). Written informed consent was obtained from all participants prior to their inclusion in the study. The study was conducted in accordance with the Declaration of Helsinki.

Patient selection

We included adult patients (aged≥18 years) consecutive detected through the PPLSH programme or by consultations and referred for evaluation of SADRs to Pharmacovigilance Unit. Eligibility criteria included a causality assessment of at least “possible,” as determined by the appropriate standardized algorithm (either the Spanish Pharmacovigilance System [SPVS] or the Roussel Uclaf Causality Assessment Method [RUCAM] for hepatitis cases), and the provision of written informed consent. Organ-specific SADRs were defined as reactions predominantly affecting a single organ system (hepatic, renal, hematologic, neurologic, or pancreatic) based on clinical, analytical, and imaging criteria, in the absence of systemic hypersensitivity manifestations such as SJS/TEN or DRESS [13]. LTT and flow cytometry analyses were performed on the patients, and the corresponding results are provided in Supplementary Table 1. Exclusion criteria included incomplete clinical data, ongoing immunosuppressive therapy, or active infectious disease at the time of testing.

A control group comprising 25 healthy individuals without any history of drug hypersensitivity was also evaluated.

Data collection and variables

Patient data were extracted from the hospital’s electronic health records (EHRs) system, HCIS. The collected variables included patient demographics (age and sex), medication history, and details of the adverse drug reaction (ADR), such as its type and interval between de reaction and study (months). Suspected drugs were then coded and grouped using the World Health Organization (WHO) Anatomical Therapeutic Chemical (ATC) classification system. The fifth level (chemical substance) was used for specific descriptions, and the second level (therapeutic main group) was used for pharmacological group analyses.

In vitro diagnostic tests

Flow cytometry assay (CD69 upregulation)

Heparinized whole blood samples were used for the flow cytometry assay. Each sample was tested in duplicate under the following conditions: (i) stimulation with the suspected drug, (ii) unstimulated negative control, and (iii) positive control with phytohemagglutinin (PHA).

For each condition, 100 µL of blood were mixed with 100 µL of RPMI-1640 medium. Drug-stimulated wells received 50 µL of the corresponding drug dilution; negative controls received 50 µL of saline solution; and PHA 20 μg/mL was added to positive controls. Cells were cultured in the presence of suspected drugs at concentrations of 1, 10 and 100 μg/mL in duplicate, and occasionally, a lower or higher concentration (0.1, 200 or 500 μg) were used. Following incubation for 72 h, samples were washed with PBS and were then stained using a master mix of monoclonal antibodies: 5 µL each of anti-CD3 FITC, anti-CD4 PerCP, and anti-CD69 PE (BioLegend. San Diego, CA, USA). A PE Mouse IgG1, κ Isotype control (Biolegend) was used to establish background fluorescence. Samples were incubated for 20–30 min at room temperature in the dark. After staining, red blood cells were selectively lysed using IOTest 3 Lysing Solution (10X, REF A07799, Beckman Coulter–Immunotech, Marseille, France), while leukocytes were preserved for subsequent analysis. Samples were then washed with PBS and immediately analyzed using a DxFlex flow cytometer (Beckman Coulter, Brea, CA, USA).

Analysis was focused on the CD3⁺CD4⁺ T lymphocyte population. The gating strategy was the following:

1. Lymphocytes were identified based on their forward and side scatter FSC/SSC properties.

2. From the lymphocyte gate, CD3⁺ T cells were selected.

3. Within the T cell population, CD4⁺ helper T cells were gated for final analysis.

The expression of CD69 was quantified on this CD3⁺CD4⁺ population. A result was considered positive if it produced an SI≥2, calculated by dividing the percentage of CD69⁺ cells in the drug-stimulated sample by the percentage in the unstimulated control. This threshold was selected for consistency with the LTT criteria and is supported based on reports demonstrating its diagnostic value in delayed drug hypersensitivity reactions [11]. A representative example of the LTT performed by flow cytometry is shown in Figure 1. A detailed list of all tested drug concentrations and results per patient is provided in Supplementary Table 1.

Figure 1:

Representative flow cytometry analysis of CD69 upregulation on CD4⁺ T cells. (A) The left panel shows the unstimulated negative control, with a basal CD69 expression of 1.20 %. The right panel shows cells treated with azithromycin condition, with CD69 expression of 1.87 %, corresponding to a stimulation index (SI) of 1.56 (1.87/1.20). (B) The left panel shows the unstimulated negative control, with a basal CD69 expression of 1.14 %. The right panel shows cells treated with azithromycin, with CD69 expression of 4.10 %, corresponding to an SI of 3.60 (4.10/1.14), clearly exceeding the positivity threshold of ≥2.

Causality assessment

Causality was assessed using the Spanish Pharmacovigilance System (SPVS) algorithm [17] and updated Roussel Uclaf Causality Assessment Method (RUCAM) for hepatitis cases [18]. These algorithmic assessments were supplemented by clinical judgment and available diagnostic data. For inclusion in the analysis, drugs were required to have at least a causality category of at least ‘possible’ (defined as a score of ≥ 4 for the SPVS algorithm or≥3 for RUCAM). Higher scores (≥6) were used in subgroup analyses to assess diagnostic sensitivity under stricter criteria.

As drug provocation testing is contraindicated in SADRs, an operational gold standard was established for our analysis. A causality score of ≥ six (on the SPVS algorithm, or RUCAM for hepatotoxicity) was used, as this threshold indicates a high probability of causality. This is a validated approach previously used in key reference studies, including work from our own group [14] and the PIElenRed group [16].

Statistical analysis

Sensitivity and specificity were calculated for each diagnostic assay, for drug class, and for the combination of both tests. For the analysis of the combined tests, a case was considered positive if a positive result was obtained in at least one of the two assays (LTT or flow cytometry). Subgroup analyses were performed based on the type of hypersensitivity reaction and the pharmacological class of the suspected drugs. The ‘gold standard’ was defined by a score of ≥6 in relevant subgroups, the specific ADR type, or the pharmacological Drug Class. Continuous variables were expressed as mean ± standard deviation (SD). Categorical variables were presented as absolute numbers or percentages. Comparisons of age and sex distribution were performed using the χ² test or Fisher’s exact test for categorical data and Student’s t-test or Mann–Whitney U test for continuous data, depending on data distribution. A two-tailed p-value <0.05 was considered statistically significant. All analyses were performed using SPSS version 21.0 (IBM Corp., Armonk, NY, USA).

Distribution of adverse drug reactions

Organ-specific reactions were predominant, with drug-induced liver injury being the most frequent. Hepatocellular hepatitis accounted for 37.3 % of cases, followed by cholestatic or mixed-pattern hepatitis (15.3 %), haematological disorders (10.7 %), nephritis (10.7 %), neurological syndromes (8.7 %), and pancreatitis (8.0 %). A miscellaneous group (9.3 %) included vasculitis, gastrointestinal, respiratory, and cutaneous syndromes (Table 1).

Suspected drugs

A total of 400 suspected drugs were identified. Systemic antibacterials were the most frequently implicated agents (29.7 %), followed by excipients (10.5 %), analgesics (10.0 %), anti-inflammatory and antirheumatic products (8.2 %), antineoplastics (6.0 %), and lipid-modifying agents (4.2 %). Other less frequent classes included antithrombotics, antiepileptics, immunosuppressants, and herbal products (Table 2).

In vitro test sensitivity by reaction type.

Sensitivity of the lymphocyte transformation test (LTT) varied across reaction types, ranging from 17 % (pancreatitis) to 47 % (haematological disorders). Flow cytometry (CD69 upregulation) showed similar variability, with sensitivities between 20 and 38 %. The combined use of both tests improved sensitivity across all categories, reaching up to 64 % in cholestatic hepatitis and 63 % in nephritis (Table 3).

In vitro test sensitivity by drug class

LTT sensitivity was highest for psychostimulants (86 %), followed by analgesics and anti-inflammatory drugs (40–50 %). Flow cytometry demonstrated higher sensitivity for lipid-lowering agents (50 %) and analgesics (50 %). Combined testing yielded the highest diagnostic performance in psychostimulants (100 %), followed by analgesics and anti-inflammatory drugs (70 %) (Table 4).

Influence of causality score on diagnostic sensitivity

Diagnostic sensitivity increased with higher causality scores. In patients with at least a causality category of “possible” (n=148), LTT and flow cytometry showed sensitivities of 46 and 30 %, respectively, and 63 % when combined. In patients with scores≥6 (n=103), sensitivity increased to 50 % for LTT, 33 % for flow cytometry, and 70 % when both tests were combined. This trend persisted in the subgroup with scores≥9 (n=23), where LTT alone reached 56 % sensitivity (Table 5).

Diagnostic performance and reaction patterns

Organ-specific SADRs were dominant in our population, with drug-induced hepatocellular hepatitis accounting for over one-third of cases. This pattern is consistent with epidemiological reports from pharmacovigilance databases, where liver injury, haematological toxicity, and nephritis are frequently represented in serious drug-induced adverse events [19], [20], [21], [22]. The diversity of reactions observed – ranging from hepatic to neurological and haematological syndromes – highlights the broad immunopathological spectrum of SADRs and underscores the need for flexible diagnostic approaches. LTT is the most widely utilized in vitro assay with a specificity of 85–100 %, whereas its sensitivity varied depending on the drug and types of ADRs ranging from 27 to 74 % [23].

The evaluation of CD69 expression has been proposed as a useful approach for identifying reactive T lymphocytes in patients with suspected drug hypersensitivity reactions [24]. In this study, CD69 expression was assessed in a cohort of 15 patients presenting with cutaneous manifestations and positive lymphocyte transformation test (LTT) results. An upregulation of CD69 on CD4⁺ T cells was observed exclusively in response to the drugs implicated in the hypersensitivity reactions [11].

We observed that LTT alone had moderate sensitivity across most reaction types (17–47 %), with higher performance in haematological and nephrotoxic presentations. Flow cytometry showed variable sensitivity (20–38 %), with a slight advantage in nephritis and pancreatitis. When both assays were combined, sensitivity improved in all categories, reaching 64 % in cholestatic hepatitis and 63 % in nephritis. These results reinforce the view that single in vitro tests may not be sufficient and that multimodal strategies increase the likelihood of correctly identifying the culprit drug [25].

Influence of drug class

Systemic antibacterials were the most frequently implicated drugs, followed by excipients, analgesics, and antineoplastic agents. This is in line with prior data showing a high burden of antibiotic-related SADRs [26], [27], [28]. Interestingly, our analysis revealed a wide variability in test sensitivity across drug classes. LTT showed notably high sensitivity (86 %) in psychoanaleptic-related reactions and 50 % in anti-inflammatory agents. Flow cytometry was particularly useful for analgesics and lipid-modifying agents, where it outperformed LTT in isolated use. When combined, the assays reached 100 % sensitivity in the psychoanaleptic group, illustrating the potential of in vitro synergy in identifying immunogenic compounds.

The observed differences may be attributable to variations in the immunogenic properties, metabolic activation, and antigen presentation pathways inherent to each drug class. Certain psychoanaleptics and anti-inflammatory agents have been reported to act as haptens or prohaptens, forming reactive metabolites or stable drug–protein conjugates that can be recognized by T cells [29], 30]. This hapten-dependent mechanism of T-cell activation is consistent with the responses detected by both proliferation assays and early activation marker expression (e.g., CD69), as demonstrated in in vitro models of drug hypersensitivity [31], 32].

Interestingly, the LTT exhibits notably low sensitivity to statins (17 %) compared with CD69 expression (50 %). This disparity is likely attributable to the established immunomodulatory and anti-inflammatory properties of statins, including the inhibition of T lymphocyte proliferation, modulation of adhesion and co-stimulatory molecules, and reduction of pro-inflammatory cytokines [33], 34]. Therefore, although statins may induce early lymphocyte activation (detectable by CD69, an early marker on T, B, and NK cells), these inherent pharmacological effects could subsequently curb cellular proliferation, the later-stage process assessed by the LTT.

Role of causality algorithms

An important observation in our study is the stepwise increase in diagnostic sensitivity with higher causality scores. In patients with a score≥6, LTT sensitivity reached 50 %, and the combined tests achieved 70 %. These findings validate the utility of integrating structured clinical assessment tools, such as the Naranjo or ALDEN algorithms, to guide interpretation of in vitro diagnostics. Previous studies have similarly shown that combining objective causality scoring with laboratory testing improves specificity and reduces the risk of misclassification [14], 35].

This is particularly relevant considering that both LTT and flow cytometry, while informative, can yield false positives (e.g., non-specific activation) or false negatives (e.g., lack of memory T-cell response). Thus, causality scores function as a valuable pre-test filter, improving post-test diagnostic confidence.

Comparison with traditional testing

The limited utility of conventional skin testing and drug provocation in SADRs is well documented. In severe cutaneous or systemic reactions, skin tests carry risks of exacerbation, and DPTs are often contraindicated due to the threat of reactivation or fatal progression [36]. Our data support the use of LTT and flow cytometry as safer, mechanistically sound alternatives that do not expose patients to harmful stimuli.

It is worth noting, however, that in vitro diagnostic tests, such as the LTT and flow cytometry-based assays, have inherent limitations. As functional cellular assays, they necessitate the use of fresh cells and strictly standardized culture conditions, which can be challenging to maintain consistently across different laboratories. Moreover, there is a lack of universally validated cut-off values for defining positive responses; for instance, stimulation index thresholds in LTT vary widely – from 1.8 to 4 – depending on the specific drug and the methodology employed. This variability is further compounded by inter-laboratory differences in protocols, including variations in drug concentrations, incubation times, and read-out systems, all of which can significantly impact test outcomes. Additionally, factors such as concomitant medications, especially immunosuppressants like corticosteroids, and the timing of the test relative to the hypersensitivity reaction, can influence the sensitivity and specificity of these assays. Collectively, these challenges limit the widespread adoption of LTT and flow cytometry-based tests in routine clinical practice [36], [37], [38].

Strengths and limitations

A major strength of our study lies in its relatively large and clinically diverse patient cohort, allowing subgroup analysis by reaction type, drug class, and causality score. Additionally, the parallel use of two independent immunological readouts provides a comprehensive functional assessment of T-cell responses. Nonetheless, the study has several limitations. The lack of confirmatory provocation testing – as ethically justified – means we relied on clinical scoring systems as the diagnostic reference standard. Similarly, while we acknowledge the lack of a universally standardized cut-off for the LTT, our choice of a stimulation index (SI) ≥ 2 is robustly supported by foundational literature and its validation in reference study groups, mitigating the variability reported in the field. Furthermore, not all drugs were tested in vitro for each patient due to technical or logistical constraints, which could underestimate the assays’ sensitivity. Finally, a methodological limitation of our study is that the flow cytometry analysis focused exclusively on the CD4⁺ helper T-cell population. While this choice was based on the central role of these cells in orchestrating most delayed-type hypersensitivity responses, we acknowledge that the simultaneous analysis of CD8⁺ cytotoxic T cells would have offered a more comprehensive characterization of the cellular response. The inclusion of the CD8⁺ population in future studies could provide valuable insights, particularly in reactions mediated by a direct cytotoxic mechanism.

Clinical implications and future directions

Our findings have clear implications for clinical practice. In settings where skin testing is unsafe or uninformative, LTT and flow cytometry offer viable, evidence-supported alternatives. The addition of flow cytometry may be particularly useful when LTT is negative or inconclusive, and vice versa. Given their complementary profiles, their combined use – especially when interpreted within the context of validated causality algorithms – should be considered a best-practice diagnostic strategy in tertiary allergy and pharmacovigilance centres.

Looking forward, the development of standardized protocols, multi-centre validation studies, and cost-effectiveness analyses will be essential to support the integration of these assays into clinical workflows. Emerging technologies, including high-throughput T-cell profiling, cytokine panels, and single-cell RNA sequencing, may further enhance diagnostic precision and open new avenues for personalized approaches in drug hypersensitivity.