The predictive value of ICI response is traditionally based on the evaluation of immunohistochemical (IHC) expression of specific proteins (i.e., PD-L1) detectable in patients’ neoplastic tissue, mostly in the advanced stages of the disease. The PD-1/PD-L1 and CTLA-4 axes are described in detail in the Supplementary Material . In the last 15 years, advanced automated techniques have been developed for the preparation of stained sections with monoclonal antibodies to minimize interpretation errors and standardize immunohistochemical analysis. However, this method is not without limitations, leading to very low predictive value and poor patient stratification. It has been demonstrated that some patients with high PD-L1 expression do not respond to ICIs. In contrast, others with low or absent PD-L1 expression may still derive benefit, highlighting the imperfect nature of this biomarker (2).

In the pursuit of more accurate and reliable methods for visualizing molecular interactions, researchers have turned to advanced imaging techniques like Fluorescence Lifetime Imaging Microscopy (FLIM) and Förster Resonance Energy Transfer (FRET). FLIM, which emerged in the late 1980s, initially focused on measuring the decay rates of fluorescent molecules, offering a novel way to study the microenvironment of these molecules beyond what was possible with traditional fluorescence intensity imaging. Over the decades, continuous technological advancements and refinements have expanded FLIM’s applications, making it a crucial tool for real-time visualization of molecular dynamics. Today, FLIM, often coupled with FRET, allows for the detailed examination of the molecular environment and interactions through autofluorescence of cellular components (e.g., NADH, collagen) or fluorescent probes labeling target molecules. These technologies provide high spatial and temporal resolution (3), enabling researchers to gain deep insights into cellular processes and molecular interactions, thus paving the way for breakthroughs in fields like cancer research and immunotherapy.

Current research has focused primarily on PD-1 and its ligand PD-L1, but new research is emerging on CTLA-4. With proper standardization of protocols, this technology may represent a reliable and effective tool for the analysis of sensitivity in candidate patients to ICI treatment.

FLIM and FRET are powerful biomedical imaging and molecular biology techniques. FLIM measures the fluorescence decay rate from excited molecules ( Figures 1 , 2 ), providing information on the local biochemical environment. On the other hand, FRET detects energy transfer between two fluorophores in close proximity, allowing for the study of molecular interactions and dynamics.

FLIM-FRET techniques, more thoroughly explained in the Supplementary Material , measure the fluorescence lifetime of the donor, avoiding contamination from the acceptor. FRET efficiency is determined by comparing the donor’s fluorescence lifetimes in the presence and absence of FRET. This method allows clear visualization of lifetime decreases in regions where FRET occurs (6, 7). The main advantages of FLIM-based FRET measurements include the ability to distinguish between interacting and non-interacting donor fractions, which is crucial for protein-interaction experiments that often involve a mix of interacting and non-interacting proteins (6, 7).

This review will provide a comprehensive analysis of the current state of FLIM technology for the qualitative and quantitative evaluation of ICI response. We will highlight the advantages and limitations of this relatively new technology based on the most relevant studies in the recent literature. Lastly, we will address current challenges and future directions for this technology. The questions that guided this review were: Is FLIM equally as effective at quantifying immune receptor expression as IHC? Can FLIM be used to stratify patients for ICI treatment better than IHC? What are its strengths and limitations?

After formulating the research question and reviewing PRISMA methodologies for systematic reviews, the team agreed on a comprehensive literature identification, screening, and documentation approach. Our research included literature focusing on predicting ICI response through FLIM written in either English or Italian. Eligible studies included retrospective cohort studies, case series, in vitro experiments, in vivo animal models, and other non-randomized observational studies. Exclusion criteria included conference abstracts, reviews, case studies, and studies lacking relevant conclusions.

The PRISMA checklist for systematic reviews was used to draft our work and ensure quality. The search was conducted on April 15, 2025, spanning several sources selected for their relevance to immunology, immunotherapy, and fluorescence imaging. Specifically, the search was performed on PubMed, Embase, and Scopus. Both reference-list scanning and grey literature research were undertaken. Everything is summarized in the flowchart ( Figure 3 ).

The following keywords and MeSH were used: “((FLIM) OR (FRET) OR (fluorescence lifetime)) AND ((PD-1) OR (PD-L1) OR (programmed death ligand 1) OR (CTLA4) OR (LAG-3) OR (TIM-3))” on PubMed, “(flim OR fret OR ‘fluorescence lifetime’) AND (‘pd 1’ OR ‘pd l1’ OR ‘programmed death ligand 1’ OR ctla4 OR ‘lag 3’ OR ‘tim 3’)” on Embase, and “(TITLE-ABS-KEY (flim) OR TITLE-ABS-KEY (fret) OR TITLE-ABS-KEY (“fluorescence lifetime”) AND TITLE-ABS-KEY (pd-1) OR TITLE-ABS-KEY (pd-l1) OR TITLE-ABS-KEY (“programmed death ligand 1”) OR TITLE-ABS-KEY (ctla-4) OR TITLE-ABS-KEY (lag-3) OR TITLE-ABS-KEY (tim-3))” on Scopus. The search yielded relevant articles published in “Analytica” by MDPI, “Research Square”, “CTM”, “Cancer Research” by AACR, “Journal of Surgical Oncology” by Wiley, and “Biophysical Chemistry”. Furthermore, reference scanning was performed on these articles. The first- and second-level screening, and data extraction were performed independently by several authors (CC, GF, ML, GM, AM, and FP) to ensure robustness and minimize errors.

Across the included studies, the overall risk of bias was moderate, Table 1 . None of the studies were RCTs, and the two human studies (8, 10) were retrospective cohorts with inherent confounding and selection biases. These observational studies did not properly control for all potential prognostic factors, which could significantly influence the observed association between FLIM-based biomarkers and ICI outcomes. However, objective endpoints have been applied (e.g., survival) and, in one case, a blinded multi-site assay analysis to mitigate measurement bias (Sánchez-Magraner et al., 2023). The in vitro proof-of-concept study was at low risk of bias, benefiting from a controlled experimental setup and objective readouts. The preclinical animal studies were generally well-conducted but still exhibited some risk of bias common to exploratory animal experiments, such as unclear blinding of outcome assessment and, in one case, non-random group assignment. One murine study did implement randomization for therapy vs. control groups, strengthening its internal validity. Overall, while all included studies had methodological limitations (e.g., retrospective design, small sample sizes, or incomplete reporting of blinding), no study was found to have a high or critical risk of bias. This suggests that the current evidence, though preliminary, is not compromised by fatal bias; still, the moderate risk-of-bias across studies underscores the need for cautious interpretation of the findings and highlights the importance of more rigorous future research (e.g., prospective trials) to confirm FLIM’s predictive value.

Currently, the standardized FDA-approved method in almost all Pathology Departments all over the globe to quantify IC expression (i.e., PD-L1) is immunohistochemistry, through approved kits with specific antibodies (i.e., 28-8, 22C3, SP263, and SP142) (14). Based on immunohistochemical analysis, compared to non-expressing subjects, patients with immune checkpoint overexpression present with a stronger antitumor activity and are more likely to benefit from ICI (13, 15, 16). However, although it is an “easy-to-use”, fast, and inexpensive method, as for other immunohistochemical evaluations, it can only provide a momentary picture of the microenvironmental status in a confined region of ex vivo specimen; furthermore, protein expression could be influenced by the concentration of fixative used or other variables related to instruments used, or the inter-observer variability on data interpretation (17, 18).

FLIM has evolved exponentially from 1988, when it was first introduced, until today; merging theoretical techniques with biomedical research. Nowadays, it can be used in combination with other imaging technologies to gather further information about the cellular microenvironment.

Recently, FLIM has gained technological advancements resulting in an improvement in the precision of analysis, as well as a broadening of this technology’s applications. These advancements have allowed it to match the precision of IHC for the analysis of ICI therapy at a microenvironmental level (8–13). However, it is unclear how responders can be differentiated from non-responders. Traditionally, IHC has been used to quantify the expression of receptors, but the expression alone does not greatly correlate to efficacy. On the contrary, functional engagement measured with FLIM between drug and receptor is a promising predictor for the success of the therapy by measuring it independently from their concentration. The clinical application would mean the inclusion of low-expressing patients who would normally be excluded from the therapy or the exclusion of non-responding high-presenting patients who would needlessly suffer the side effects.

FLIM and FRET are reliable imaging techniques that could overcome some of the limitations of IHC. Table 3 shows a direct comparison between the technical and practical capabilities of IHC and FLIM. While IHC is the gold standard for predicting the body’s response to various ICIs, its limitations have become increasingly evident (2). Despite the scarce research on this topic, the usage of FLIM and FRET to test ICI efficacy have shown promising results by providing real-time, non-invasive insights into molecular interactions such as PD-1/PD-L1 engagement, surpassing the static and limited biopsy samples used in IHC. FLIM detected treatment-induced changes in tumors in vivo just 2 days post-treatment, which is earlier than detectable changes in tumor volume (31). Furthermore, FLIM-FRET allows for the visualization of checkpoint interactions at a microscopic level, providing crucial information about the functional state of these molecules as shown by iFRET which detected significant interaction states in patients who were PD-L1 negative according to IHC (8). FLIM can help detect cells’ in vivo metabolism with no phototoxicity and in real-time. FLIM’s capability to quantify functional interactions offers a more comprehensive approach, potentially improving the stratification of patients for immunotherapy and reducing the ambiguity associated with IHC-based assays. It is important to underline that FLIM finds its greatest potential in vivo and relies on the fact that its results, unlike traditional fluorescence microscopies, are not dependent on the change in fluorescence intensity but on the lifetime. FLIM, with or without FRET, still suffers from major drawbacks that vary depending on its specific application. The use of FLIM technology in the analysis of ex vivo samples proves it is non-superior to more widespread methods since metabolic microenvironment characteristics are lost in the transition from the in vivo to the ex vivo, limiting its ability to provide accurate insights into molecular interactions.

Hence, the true power of FLIM resides in its in vivo application. One viable way to adopt FLIM in an ex vivo context would be through FLIM used with Raman spectroscopy (32) as the tissue cryosections partially maintain the in vivo microenvironment. However, FLIM-FRET detection probes are unsuitable when using this technique due to the need for a non-frozen tissue for their employment rendering this technique limited. The probes are a major issue even in the in vivo applications of FLIM-FRET since they are scarce in number and the existing ones yet unsuitable for human patients.

FLIM also presents some minor challenges that further research can work to resolve. For example, since the lifetime duration computation is based on statistics, the higher the number of iterations, the higher the precision and certainty of measurements. This implies that a drastic increase in time is required to obtain robust values, especially in multiplexing applications (33). Notably, scalability for clinical use is being addressed by innovations such as GPU-accelerated high-speed FLIM, which significantly reduces imaging and processing times (25). Deep learning approaches, such as Phasor U-Net, automate and accelerate lifetime extraction and multiplexing, minimizing manual intervention and enabling rapid, accurate analysis even with limited photon counts (34). High-throughput acquisition systems using array detectors and parallelized photon counting further increase scalability (35).

Another problem of FLIM in vivo applicability is the depth of measurements (31). Until 2024, the number of tumors analyzed in vivo with this technique is minimal (mainly melanomas due to their easily accessible location). Tissue depth penetration is limited as with all optical imaging modalities, and with two-photon FLIM it is around 100-130 μm (36), in vivo visualization of deep tissues (i.e., intestine, kidneys, liver) is currently limited to invasive approaches.

Since FLIM-FRET works with standard confocal microscopy, the maximum resolution possible is 200–250 nanometers (37), which is enough to make it a valuable option in studying molecular interactions and changes in the microenvironment. However, it is not comparable to other microscopy techniques (i.e., electron microscopy) although they have other serious drawbacks such as phototoxicity.

An additional challenge associated with FLIM-FRET is the complexity and the technical expertise required for its implementation. It requires strict protocols to reduce interference from extrinsic factors (i.e., pH, temperature, etc.) that may affect fluorescence decay times (31) and a profound knowledge of the biological environments and pathways involved.

Despite these barriers, commercial development is underway, companies such as JenLab (38–40) are offering FLIM-based devices for dermatological applications, indicating momentum towards clinical implementation.

Also, the cost and resource intensity of FLIM-FRET systems pose a significant barrier; the advanced imaging equipment required is expensive and often requires specialized maintenance. The economic constraint can hinder the broader adoption of FLIM-FRET in clinical practice, despite its potential benefits.

The utilization of FLIM and FLIM/FRET is expected to have a great impact on future clinical practice (41, 42) considering the effect on cancer patients’ diagnosis given the high specificity and sensitivity of the technique, particularly regarding genetically encoded biosensors reviewed by Vu et al. (41). In 2023 it has been shown that high precision and accuracy (respectively closeness of known values among them and closeness of known values with the true one) guarantee a delved and highly specific landscape of the cellular metabolism and molecular interactions previously presented in the ICI section.

This novel technology encompasses the current trend of personalized precision medicine. We are gradually diving into having a treatment specific to each patient for high-quality care and patient management.

The current state-of-art of FLIM technology strongly suggests that intraoperative guidance use of FLIM has been emerging as a relevant and consistent future application of the mentioned technology. In this setting, FLIM is invasive, as it requires direct access to tissue during surgical procedures, but it enables real-time imaging capabilities (42). At the same time, non-invasive applications are also advancing, particularly in dermatology (38–40). These highlight that there is still an optimal margin to further enhance this microscopy in both domains.

Practical pipeline development for broader clinical integration of FLIM involves creating comprehensive training programs for laboratory technologists and pathologists, deploying automated, ready-to-use FLIM systems with standardized protocols, and embedding FLIM modules into existing histopathology and cytometry platforms. Open-source toolkits like FLIMJ facilitate integration with established image analysis workflows, reducing the barrier for adoption in clinical laboratories (43). Regulatory pathway development and multi-institutional validation studies are essential for clinical acceptance.

Overall, the topic remains mostly underexplored, and further research is needed to better understand the potential of known and alternative immune checkpoint pathways such as CTLA-4, TIM-3, and LAG-3.