⁶⁸Ga-grazytracer PET for noninvasive assessment of response to immunotherapy in solid tumors and lymphomas: a phase 1/2 clinical trial

Immunotherapies such as immune checkpoint inhibition and chimeric antigen receptor (CAR)-T cell therapy have revolutionized tumor therapy. However, only a subgroup of patients respond to immunotherapy. In contrast to biopsy-based methods such as immunohistochemistry (IHC) staining of programmed death-ligand 1 (PD-L1) and measurement of tumor mutation burden¹ and immunophenotypes², imaging tools enable a systemic evaluation of tumor responses to immunotherapy in a noninvasive and longitudinal manner. The Response Evaluation Criteria in Solid Tumors (RECIST) guidelines³ and lymphoma-adapted RECIST⁴ based on computed tomography (CT) have been widely used to assess the efficacy of immunotherapy in treating solid tumors and lymphomas. However, these anatomical image-based evaluation criteria rely solely on changes in tumor lesion size and cannot discern abnormal progression patterns, especially pseudoprogression, during immunotherapy^5,6. Although immune RECIST (iRECIST)⁷, immune-related RECIST (irRECIST)⁸, and the refined Lymphoma Response to Immunomodulatory Therapy Criteria (LYRIC)⁹ optimize the timing of decision-making by including follow-up scans, these wait-and-see strategies may delay the optimal timing for adjusting treatment strategies in patients.

Considering that metabolic changes in tumors precede anatomical size alterations¹⁰, there is an increasing clinical reliance on ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography (PET) as the current standard for evaluating immunotherapy effectiveness in treating solid tumors, following the PET-Response Criteria in Solid Tumors (PERCIST)¹¹ guidelines¹². For lymphomas, the Lugano classification also includes ¹⁸F-FDG-based evaluation methods¹³. However, ¹⁸F-FDG uptake is not tumor-specific, and immunotherapy-induced tumor-infiltrating immune cells exhibit heightened glucose metabolism, potentially resulting in pseudoprogression on early ¹⁸F-FDG PET^14,15.

PET imaging of crucial immune checkpoints, such as programmed cell death protein 1 (PD-1) and its ligand PD-L1, through the use of radiolabeled antibodies or peptides, has been utilized to evaluate the effectiveness of immunotherapy in patients with non-small cell lung cancer and melanoma^16,17,18. However, the predictive utility of PD-1/PD-L1 is confined predominantly to anti-PD-1/PD-L1 therapies and offers limited insight into the assessment of other immunotherapeutic approaches^19,20. Radiotracers targeting other immune checkpoints, such as T cell immunoreceptor with Ig and ITIM domains (TIGIT)²¹ and lymphocyte-activation gene 3 (LAG-3)²² have been employed to evaluate patient responses to anti-TIGIT and anti-LAG-3 therapies. However, the clinical application of antibodies against TIGIT and LAG-3 is still in its infancy, which limits the widespread clinical utilization of the corresponding radiotracers²³. Owing to the critical role of CD8⁺ T cells in immunotherapy, CD8-targeted PET imaging has been used to predict immune responses. Tumor uptake of CD8-specific radiotracers before treatment positively correlated with patient prognosis in clinical trials^24,25. However, PET of CD8⁺ T cells has two limitations for monitoring immunotherapy. First, only a fraction of CD8⁺ T cells (tumor-associated antigen-specific populations) within the tumor microenvironment can recognize and eliminate tumor cells, making it challenging for CD8-targeted PET to identify antigen-specific CD8⁺ T cells. Second, CD8-targeted PET cannot distinguish between effector and naïve CD8⁺ T cell subsets²⁶. A retrospective analysis based on the Cancer Genome Atlas also indicated that the overall expression levels of CD8 could not predict immune responses in epithelioid mesothelioma²⁷.

To overcome the limitations associated with CD8-targeted PET imaging, radiotracers for the PET of CD8⁺ T cell effector function have been extensively investigated. For instance, radiotracers that target granzyme B, a crucial molecule secreted by activated T cells for executing effector functions, have been shown to predict early tumor responses to immunotherapy in preclinical studies^28,29. We recently developed a peptidomimetic-based PET radiotracer named ⁶⁸Ga-grazytracer, which demonstrated favorable stability in vivo, thus enabling PET imaging of granzyme B expression in patients³⁰.

In this work, we conduct a phase 1/2 investigator-initiated clinical trial (NCT05000372) to further assess the clinical utility of ⁶⁸Ga-grazytracer PET imaging. We perform ⁶⁸Ga-grazytracer PET imaging on patients with solid tumors after receiving 2–3 cycles of immunotherapies and on patients with lymphomas one month post CAR-T cell therapy. We evaluate the safety and pharmacokinetic properties of ⁶⁸Ga-grazytracer, as well as the association between ⁶⁸Ga-grazytracer tumor uptake and patient clinical outcomes.

Patient characteristics

Between September 2021 and March 2023, 24 participants (4 women and 20 men; median age, 64 years; interquartile range, 57–70 years) with malignant tumors (21 patients with solid tumors and 3 with lymphomas) who were scheduled to receive immunotherapy were enrolled (see Supplementary Fig. 1 for the consort diagram) in this prospective study (NCT05000372). The primary endpoints of this trial included safety, biodistribution, granzyme B specificity, and the predictive value of ⁶⁸Ga-grazytracer, while the secondary endpoint focused on the correlation between the uptake of ⁶⁸Ga-grazytracer and the tumor immune phenotype. One participant received durvalumab alone, twenty received anti-PD-1/anti-PD-L1 antibodies with chemotherapy, and three received CAR-T cell therapy (Table 1). All participants underwent pre-treatment ¹⁸F-FDG PET/CT, post-treatment ¹⁸F-FDG PET/CT, and ⁶⁸Ga-grazytracer PET/CT after 2–3 cycles of immunotherapies or 1 month of CAR-T cell therapy for prognostic analysis. The median follow-up time for the cohort was 23 months (15–27 months). At the one-year follow-up, 14 patients experienced disease progression, resulting in a one-year progression-free survival (PFS) rate of 41.7%. At the end of the follow-up, none of the patients had died. The median PFS for all patients was 9 months (4–21 months).

Safety and pharmacokinetics of ⁶⁸Ga-grazytracer

All patients underwent ⁶⁸Ga-grazytracer PET/CT without complications, and no drug-related side effects or other adverse physiological responses were observed. The patients did not report any abnormal symptoms, demonstrating an overall tolerance to the procedures.

We assessed the in vivo metabolic kinetics of ⁶⁸Ga-grazytracer using dynamic PET/CT (Fig. 1a). In the early phases of imaging, the peak physiological uptake of the radiotracer was evident in the kidneys, accompanied by discernible but moderate uptake in the heart, spleen, and liver. Subsequently, a significant temporal reduction in ⁶⁸Ga-grazytracer uptake within these organs was observed (Fig. 1b, c), indicating rapid renal clearance of the radiotracer. This phenomenon aligns with efficient renal metabolism and the elimination of ⁶⁸Ga-grazytracer from systemic circulation.

a Representative PET images of ⁶⁸Ga-grazytracer at 5, 10, 20, 40, and 90 min post-injection. b, c Quantified mean standard uptake value (SUV_mean) of ⁶⁸Ga-grazytracer in healthy tissues at 5, 10, 20, 40, and 90 min post-injection (n = 4, collected from 4 different patients). The curves were fitted and depicted using a quadratic (second-degree polynomial) model. Data points represent the median SUV_mean values, with error bars representing the interquartile range. Source data are provided as a Source Data file.

Full size image

Tumor uptake of ⁶⁸Ga-grazytracer

We analyzed the temporal uptake dynamics of ⁶⁸Ga-grazytracer in tumors and healthy organs within 40 min of intravenous administration (Fig. 2a; Supplementary Fig. 2a–d). General additive mixed models were utilized to describe the biodistribution of ⁶⁸Ga-grazytracer in healthy organs (Supplementary Table 1). Owing to the rapid clearance of ⁶⁸Ga-grazytracer, its tumor uptake steadily decreased over time (Fig. 2b). However, the tumor-to-liver (Fig. 2c) and tumor-to-blood (Supplementary Fig. 2e) uptake ratios exhibited an upward trend, peaking at approximately 30 min post-injection. In the PET images of patients, diverse individual patterns of ⁶⁸Ga-grazytracer uptake were noted, with a subset exhibiting tumor-rim uptake (Fig. 2d). Additionally, a significant uptake of ⁶⁸Ga-grazytracer was observed in some metastatic lesions (Fig. 2e), suggesting its potential to assess whole-body granzyme B expression.

a Representative PET images of ⁶⁸Ga-grazytracer at 2, 5, 10, 20, 30, and 40 min post-injection. b Quantified maximum standardized uptake value (SUV_max) of ⁶⁸Ga-grazytracer uptake in tumors at 2, 5, 10, 20, 30, and 40 min post-injection (n = 4, collected from 4 different patients). Data points represent the median SUV_max values, with error bars representing the interquartile range. c Calculated mean SUV_max of the tumor-to-liver ratio at different time points (n = 4, collected from 4 different patients). Data points represent the median calculated tumor-to-liver ratio values, with error bars representing the interquartile range. d Representative PET/CT images of ⁶⁸Ga-grazytracer depicting the tumor uptake in the rim. e Representative PET/CT images of ⁶⁸Ga-grazytracer uptake in a metastatic lesion. (i) Uptake in a brain metastasis (SUV_max of 1.9) of a patient with right lung adenocarcinoma. (ii) Uptake in a pleural metastasis (SUV_max of 3.7) of a patient with left lung adenocarcinoma. f Immunohistochemical staining of granzyme B (GrzmB) in the tumor tissues harvested from three patients with different ⁶⁸Ga-grazytracer uptake (SUV_max values of 2.0, 2.7, and 4.1). Experiments were conducted independently three times with similar results. Scale, 50 μm. g Correlation between the tumor uptake of ⁶⁸Ga-grazytracer and the quantified GrzmB levels (n = 7, r = 0.7701, P = 0.0094). h SUV_max of ⁶⁸Ga-grazytracer in tumor lesions with “desert” (n = 11) and “non-desert” immune phenotypes (n = 7). Data are presented as median with interquartile range. Tumors are indicated by arrows in the PET and PET/CT images. Two-sided P value was calculated using Pearson’s correlation analysis (g) and Mann‒Whitney U test (h). Source data are provided as a Source Data file.

Full size image

To further elucidate the in vivo specificity of ⁶⁸Ga-grazytracer, we performed IHC staining for granzyme B on surgical specimens from 7 patients who were eligible for surgery following immunotherapy (Fig. 2f). The results revealed a significant correlation between tumor uptake of ⁶⁸Ga-grazytracer and the expression levels of granzyme B in tumor tissues (Pearson’s r = 0.7701, P = 0.0094; Fig. 2g). Using the CD8 staining results from tumor tissue samples collected from patients prior to immunotherapy (Supplementary Fig. 3a, b), we categorized the tumor immune phenotypes into immune “desert” (low CD8 expression) and immune “non-desert” (high CD8 expression). The maximum standardized uptake value (SUV_max) of ⁶⁸Ga-grazytracer was significantly higher in the “non-desert” immune phenotype of tumors than in the immune “desert” tumors (P = 0.0017; Fig. 2h), suggesting the role of ⁶⁸Ga-grazytracer PET in noninvasively distinguishing the immune phenotypes of tumors in patients.

Predictive values of ⁶⁸Ga-grazytracer PET uptake for immunotherapy outcomes

Using CT data from patient lesions 6 months post-treatment, we evaluated the overall response using the RECIST 1.1 and Lugano criteria. Complete response (CR) and partial response (PR) were observed in 45.8% (11/24) of the patients, whereas progressive disease (PD) and stable disease (SD) were observed in 54.2% (13/24) of the patients. We further analyzed and compared the predictive accuracy of pre-treatment PD-L1 IHC with three imaging parameters: the SUV_max of ⁶⁸Ga-grazytracer after immunotherapies, changes in the Sum of the Longest Diameter of CT after therapies compared to baseline (ΔSLD), and changes in SUV_max (ΔSUV_max) of ¹⁸F-FDG after therapies compared with baseline. Discordance between PD-L1 IHC results and patient prognosis (PD or SD vs. PR or CR) was observed (Fig. 3a), suggesting the limited predictive value of traditional biopsy-based pathological assessment for immunotherapy. In patients who achieved PR or CR according to RECIST 1.1 or Lugano criteria, the SUV_max of ⁶⁸Ga-grazytracer was significantly higher than that in the patients with PD or SD (P = 0.0004; Fig. 3b). By contrast, the ΔSLD of CT scan (P = 0.28) and ΔSUV_max of ¹⁸F-FDG (P = 0.13) showed a tendency to differ between these two groups, but these differences did not reach statistical significance (Fig. 3c, d). Supplementary Fig. 4a–c further illustrates the differences in the aforementioned imaging parameters among patients with varying overall responses. Receiver operating characteristic (ROC) curve analysis further confirmed the superior predictive performance of the SUV_max of ⁶⁸Ga-grazytracer (area under the ROC curve [AUC] = 0.899; 95% confidence interval [CI], 0.77–1.00; P = 0.0010) than that of the ΔSLD of CT scan (AUC = 0.636; 95% CI, 0.41–0.86; P = 0.26) and ΔSUV_max of ¹⁸F-FDG (AUC = 0.685; 95% CI, 0.46–0.91; P = 0.12) (Fig. 3e).

Comparison of the positive PD-L1 proportion (a), SUV_max of ⁶⁸Ga-grazytracer (b), ΔSLD of CT scan (c), and ΔSUV_max of ¹⁸F-FDG (d) in patients categorized as PD/SD (n = 13) and PR/CR (n = 11) according to the overall response. e Receiver operating characteristic (ROC) curves for the SUV_max of ⁶⁸Ga-grazytracer, ΔSLD of CT scan, and ΔSUV_max of ¹⁸F-FDG (n = 24 for each group). The area under the curve (AUC) was calculated for each ROC curve, with 95% confidence intervals derived using the Wilson/Brown method. Statistical differences in AUCs were evaluated using DeLong’s tests, resulting in two-sided P values. To account for the false discovery rate in multiple comparisons, the Bonferroni correction was utilized to adjust the P values. Two-sided P value was determined by the Chi-square and Fisher’s exact test (a) and Mann–Whitney U test (b–d). SUV_max the maximum standardized uptake value, ΔSLD changes in the sum of the longest diameter, ΔSUV_max changes in the maximum standardized uptake value, PD progressive disease, SD stable disease, PR partial response, CR complete response. Source data are provided as a Source Data file.

Full size image

To validate the clinical utility of ⁶⁸Ga-grazytracer in assessing immunotherapy efficacy, patients were stratified based on the median SUV_max of ⁶⁸Ga-grazytracer, and its predictive accuracy was compared with that of the existing clinical criteria, including RECIST 1.1, PERCIST, and Lugano. The highest prediction value for the overall response 6 months post-treatment was observed in the group assessed with ⁶⁸Ga-grazytracer PET (SUV_max cutoff of 2.3) (P = 0.0022; Fig. 4a) than those with RECIST 1.1/Lugano (based on ΔSLD of CT scan) (P = 0.10; Fig. 4b) and PERCIST/Lugano (based on the changes in peak SUV corrected for lean body mass of ¹⁸F-FDG [ΔSUL_peak] or Deauville score of the SUV_max of ¹⁸F-FDG) (P = 0.12; Fig. 4c).

a–c The Sankey diagram compares the concordance between early assessment results based on ⁶⁸Ga-grazytracer uptake (SUV_max cutoff of 2.3) (a), ΔSLD of CT scan (RECIST 1.1 and Lugano) (b), and ΔSUV_max of ¹⁸F-FDG (PERCIST and Lugano) (c) and the overall response based on RECIST 1.1 or Lugano 6 months post-treatment. Two-sided P value was determined by Chi-square and Fisher’s exact tests (a–c). d–f Progression-free survival (PFS) curves estimated using the Kaplan–Meier method according to the early assessment results based on ⁶⁸Ga-grazytracer uptake (d), ΔSLD of CT scan (RECIST 1.1 and Lugano) (e), and ΔSUV_max of ¹⁸F-FDG (PERCIST and Lugano) (f). The log-rank test was used to compare PFS curves across different groups, and Cox regression models were employed to estimate the hazard ratios. In this context, “time 0” refers to the commencement of treatment, serving as the starting point for measuring the duration of PFS (d–f). SUV_max the maximum standardized uptake value, RECIST Response Evaluation Criteria in Solid Tumors, PERCIST PET-Response Criteria in Solid Tumors, PR partial response, CR complete response, PD progressive disease, SD stable disease, PMR partial metabolic response, CMR complete metabolic response, PMD progressive metabolic disease, SMD stable metabolic disease. Source data are provided as a Source Data file.

Full size image

In the follow-up studies, patients with an SUV_max of ⁶⁸Ga-grazytracer >2.3 displayed significantly higher PFS than did patients with an SUV_max of ⁶⁸Ga-grazytracer ≤2.3 (P = 0.0009; hazard ratio, 7.3; 95% CI, 2.7–20.2; Fig. 4d). In contrast, the defined responders (PR and CR) by RECIST 1.1/Lugano showed no superior PFS compared with that of defined nonresponders (P = 0.80; hazard ratio, 1.1; 95% CI, 0.4–3.5; Fig. 4e). Patients assessed as the responders (partial metabolic response [PMR] and complete metabolic response [CMR]) using PERCIST/Lugano showed a trend towards improved PFS but no significant differences were observed (P = 0.11; hazard ratio, 2.2; 95% CI, 0.8–6.6; Fig. 4f). Proportional hazard assumptions were met for the Cox model and demonstrated that the impact of ⁶⁸Ga-grazytracer uptake on the risk of disease progression did not significantly vary over time (Supplementary Table 2). Cox regression analysis also revealed that ⁶⁸Ga-grazytracer uptake was significantly associated with PFS after adjusting the confounding factors (Supplementary Table 3). Therefore, the criteria using SUV_max cutoff of 2.3 for ⁶⁸Ga-grazytracer PET demonstrated superior accuracy in predicting the responsiveness to immunotherapy compared to RECIST 1.1/Lugano and PERCIST/Lugano. Additionally, we performed a PFS analysis in patients with solid tumors alone and found similar results (Supplementary Fig. 5).

In a representative case of tumor pseudoprogression, a patient with lung adenocarcinoma, identified by a 5.2 × 3.3 cm right lung mass, had a SUV_max of 7.6 on baseline ¹⁸F-FDG PET (Fig. 5a). After two cycles of immunotherapy and chemotherapy, no changes in tumor size were observed; however, the tumor showed a flare-up on ¹⁸F-FDG PET, with the SUV_max increasing to 11.3 (Fig. 5b). ⁶⁸Ga-grazytracer PET showed high tumor uptake with a SUV_max of 5.2 (Fig. 5c), suggesting CD8⁺ T cell infiltration and activation within the tumor. A concurrent chest CT scan performed alongside the ⁶⁸Ga-grazytracer PET showed no significant change in tumor size (Fig. 5d). However, the CT results obtained 19 days later (Fig. 5e) revealed a notable decrease in tumor size. This not only indicates the therapeutic efficacy in this patient but also highlights the predictive power of ⁶⁸Ga-grazytracer PET in assessing immunotherapy responses earlier than conventional CT scanning.

a The baseline ¹⁸F-FDG PET/CT revealed a mass (5.2 × 3.3 cm in size) with intense ¹⁸F-FDG uptake (SUV_max of 7.6) in the upper lobe of the right lung. b ¹⁸F-FDG PET/CT after two cycles of treatment (immunotherapy and chemotherapy) showed no changes in tumor size, while the ¹⁸F-FDG uptake was increased compared with baseline (SUV_max of 11.3 vs. 7.6). c ⁶⁸Ga-grazytracer PET/CT after two cycles of treatment showed a mass in the upper lobe of the right lung with inhomogeneous increased activity (arrows; SUV_max of 5.2). d Chest CT after 3 days of ⁶⁸Ga-grazytracer PET/CT revealed no change in the tumor size. e Chest CT after 19 days of ⁶⁸Ga-grazytracer PET/CT indicated a significant reduction in tumor size. Arrows indicate tumors in the PET, PET/CT, and CT images.

Full size image

Two cases showing a correlation between ⁶⁸Ga-grazytracer PET and long-term prognosis in patients with lymphomas who received CAR-T cell therapy are shown in Fig. 6. Patient 1 was diagnosed with follicular lymphoma (a left abdominal mass of 12.1 × 8.7 cm) with an SUV_max of 13.9 on baseline ¹⁸F-FDG PET (Fig. 6a). After 1 month of CAR-T cell therapy, ⁶⁸Ga-grazytracer PET revealed high uptake in the lymphoma (SUV_max, 5.0; Fig. 6b), suggesting a favorable outcome. Meanwhile, ¹⁸F-FDG PET/CT revealed low lymphoma uptake (SUV_max of 2.2) and reduced tumor size (7.0 × 3.1 cm) (Fig. 6c). After 14 months of CAR-T cell therapy, the patient exhibited a complete metabolic response, as determined via ¹⁸F-FDG PET (Fig. 6d). Patient 2 was diagnosed with B-cell lymphoma at the left side of the neck who had an intense activity on baseline ¹⁸F-FDG PET/CT (SUV_max of 15.9; Fig. 6e). After 1 month of CAR-T cell therapy, ⁶⁸Ga-grazytracer PET revealed no activity in the neck mass (Fig. 6f), suggesting a poor prognosis. Although ¹⁸F-FDG PET/CT revealed a reduced tumor size (2.8 × 2.4 cm) and decreased activity (SUV_max of 4.3) (Fig. 6g), the patient experienced overall disease progression after 3 months (Fig. 6h).

a–d A patient with follicular lymphoma. a The baseline ¹⁸F-FDG PET demonstrated intense ¹⁸F-FDG-avid mass (SUV_max of 13.9; 12.1 × 8.7 cm in size) in the left abdominal cavity. b ⁶⁸Ga-grazytracer PET after 1 month of CAR-T cell therapy revealed increased lesion uptake (SUV_max of 5.0). c ¹⁸F-FDG PET after 1 month of CAR-T cell therapy showed that the tumor size decreased (7.0 × 3.1 cm vs. 12.1 × 8.7 cm) and SUV_max of ¹⁸F-FDG also decreased (SUV_max of 2.2 vs. 13.9) compared with the baseline ¹⁸F-FDG PET. d ¹⁸F-FDG PET after 14 months of CAR-T cell therapy revealed a complete metabolic response in this patient. e–h A patient with B-cell lymphoma. e The baseline ¹⁸F-FDG PET showed a mass with intense ¹⁸F-FDG uptake in the left side of the neck (SUV_max of 15.9; 4.1 × 3.4 cm in size). f ⁶⁸Ga-grazytracer PET after 1 month of CAR-T cell therapy showed no radiotracer uptake in the mass at the left side of the neck. g ¹⁸F-FDG PET after 1 month of CAR-T cell therapy revealed that the tumor size decreased (2.8 × 2.4 cm vs. 4.1 × 3.4 cm) and SUV_max of ¹⁸F-FDG also decreased (SUV_max of 4.3 vs. 15.9) compared with the baseline ¹⁸F-FDG PET. h ¹⁸F-FDG PET after 3 months of CAR-T cell therapy revealed overall disease progression in this patient. Arrows indicate tumors in the PET images.

Full size image

Although CD8⁺ T cells play a pivotal role in immunotherapy, only a small subset of CD8⁺ T cells within the tumor possesses the effector function of recognizing and eliminating cancer cells³¹. Therefore, visualizing effector CD8⁺ T cells in vivo is crucial in accurately assessing the efficacy of immunotherapy. In this phase 1/2 clinical trial, we used ⁶⁸Ga-grazytracer to assess granzyme B expression levels within tumor lesions, thereby characterizing the functional status of CD8⁺ T cells post-immunotherapy. We evaluated the pharmacokinetics of ⁶⁸Ga-grazytracer in vivo, demonstrating its significant uptake in the blood pool and kidneys owing to its rapid clearance as a small molecule³². The radiotracer exhibited concentrated uptake at the tumor periphery, consistent with the tumor rim pattern of T-cell infiltration³³. Moreover, the IHC results for granzyme B in tumor specimens correlated with the tumor uptake of ⁶⁸Ga-grazytracer, confirming its specificity for granzyme B. Compared with conventional clinical assessments, such as CT, ¹⁸F-FDG PET, and PD-L1 IHC, ⁶⁸Ga-grazytracer PET more effectively predicted short-term prognosis and PFS in patients undergoing immunotherapy.

The precision of patient categorization using ⁶⁸Ga-grazytracer PET/CT during early immunotherapy significantly outperformed current clinical evaluations based on CT and ¹⁸F-FDG PET/CT (such as RECIST 1.1, PERCIST, and the 2014 Lugano criteria) in predicting short-term outcomes and overall prognosis for immunotherapy. ⁶⁸Ga-grazytracer PET/CT superiority stems from the foundational mechanism of immunotherapy, which involves reinstating potent cytotoxic T-cell responses against tumors, distinct from conventional chemotherapy and targeted therapy modalities³⁴. Using anatomical criteria (RECIST 1.1) for assessing the tumor response to immunotherapy is challenging, as alterations in tumor size may not promptly or reliably signify the tumor response to immunotherapy, with metabolic shifts and necrosis often preceding changes in tumor size³⁵. In addition, an increase in tumor size and the appearance of new lesions do not invariably indicate disease progression but could reflect immune cell infiltration³⁶. The underlying principle of ¹⁸F-FDG PET/CT-based assessments for evaluating immunotherapy efficacy relies on the close association between ¹⁸F-FDG uptake and cancer cell abundance. However, distinguishing tumors cells from infiltrating T lymphocytes is challenging because both express high-affinity glucose transporter-1 that facilitate glycolysis³⁷.

⁶⁸Ga-grazytracer PET/CT holds considerable promise across various clinical applications for evaluating the efficacy and prognosis of immunotherapies in patients with cancer. Its exceptional utility in gauging immunotherapy efficacy encompasses several dimensions. First, ⁶⁸Ga-grazytracer PET/CT facilitates the specific recognition of activated CD8⁺ T cells, aiding in discriminating pseudoprogression. This capability is instrumental for accurately evaluating immunotherapy efficacy and guiding timely therapeutic interventions. Timely recognition and continuation of therapy are pivotal in managing pseudoprogression³⁸. Second, ⁶⁸Ga-grazytracer PET/CT has emerged as a valuable tool for predicting immunotherapy efficacy during the initial treatment phases, thereby facilitating the optimization of strategy refinement. Given the heterogeneous responses observed among patients, early differentiation between responders and nonresponders is important³⁹, as it enables the timely adjustment of treatment strategies for nonresponders. Third, ⁶⁸Ga-grazytracer may have prognostic significance in patients treated with CAR-T cell therapy. Patients with high ⁶⁸Ga-grazytracer uptake in lesions may have a better long-term prognosis after CAR-T cell therapy, suggesting that ⁶⁸Ga-grazytracer PET/CT holds promise for addressing the limitations of ¹⁸F-FDG PET/CT in the early monitoring of CAR-T cell therapy. Early ¹⁸F-FDG PET/CT assessments a month post CAR-T cell therapy often fail to predict long-term responses⁴⁰. Therefore, ⁶⁸Ga-grazytracer PET/CT offers a promising approach for assessing the efficacy of CAR-T cell therapy at an early stage.

In this clinical trial, we did not perform baseline ⁶⁸Ga-grazytracer PET imaging to directly observe the changes in granzyme B levels following immunotherapy. Our previous animal studies have shown that the tumor uptake of ⁶⁸Ga-grazytracer in the absence of immunotherapy was consistently low (usually less than 1%ID/g) with negligible variability between individuals³⁰. This is because the activation of CD8⁺ T cells and secretion of granzyme B are minimal before immunotherapy. It could be inferred that the granzyme B levels in patients prior to immunotherapy are at background levels. Therefore, the SUV_max of ⁶⁸Ga-grazytracer post-immunotherapy as we used in this study could serve as an indicator of granzyme B expression induced by immunotherapy.

Due to the variety of tumor types encountered in this study, the small number of patients with specific tumor types may preclude the definition of T-cell functional status within patient subgroups. In certain patients, we observed significant therapeutic outcomes despite the presence of low ⁶⁸Ga-grazytracer uptake in the tumor area. In these cases, it is possible that granzyme B may have already penetrated tumor cells, a scenario that cannot be directly detectable by ⁶⁸Ga-grazytracer PET, which primarily detects granzyme B in the tumor microenvironment rather than within the tumor cells⁴¹. We chose to conduct granzyme B PET imaging after 2–3 cycles of immunotherapy in patients with solid tumors and one month following CAR-T cell therapy for lymphoma patients, based on our pilot study that indicated consistent ⁶⁸Ga-grazytracer uptake at one, three, and four cycles of immunotherapy. Nevertheless, dynamic consecutive imaging in patients to establish the most suitable timing for granzyme B PET imaging would be appropriate. This method has been successfully utilized in our previous animal studies³⁰ and could facilitate a more precise capture of the peak secretion phase of granzyme B.

The participants in this study include patients who received a combination of immunotherapy and chemotherapy. While studying a more homogenous group of patients undergoing immunotherapy alone could lead to stronger clinical conclusions about the utility of ⁶⁸Ga-grazytracer PET in monitoring the efficacy of immunotherapy, in clinical practice, the combination therapy is more commonly used due to several practical factors. In our previous preclinical study³⁰, we performed an extensive evaluation of ⁶⁸Ga-grazytracer in mouse models that were treated with immunotherapy alone, which encompassed both immune checkpoint inhibitors and adoptive cell transfer therapies. These studies highlighted the potential of ⁶⁸Ga-grazytracer PET for the early prediction of immunotherapy responses and for differentiating between pseudoprogression and true tumor progression. Furthermore, there is a growing body of evidence suggesting that chemotherapy can augment the effectiveness of immunotherapy via diverse mechanisms^42,43. Nonetheless, the secretion of granzyme B represents a terminal event in various antitumor immune pathways and serves as a direct measure of the cytotoxic capacity of T cells to eradicate tumor cells. This is true whether the treatment approach consists of immunotherapy by itself or in conjunction with other therapeutic modalities. As such, ⁶⁸Ga-grazytracer PET offers the potential to visualize the terminal antitumor effects in patients who have received a combination of chemotherapy and immunotherapy.

In conclusion, this study demonstrated that ⁶⁸Ga-grazytracer PET can specifically identify granzyme B expression in humans, offering a noninvasive method for monitoring the effector function of CD8⁺ T cells. Compared to conventional imaging techniques such as ¹⁸F-FDG PET and CT, ⁶⁸Ga-grazytracer PET could represent an alternative tool for predicting early-stage responses of patients with tumors to immunotherapy.

Study participants

This prospective investigator-initiated trial (NCT05000372) was approved by the Clinical Research Ethics Committee of the Peking University Cancer Hospital (no. 2021KT86) in August, 2021. The clinical study was conducted in accordance with the criteria set by the Declaration of Helsinki and all pertinent regulations concerning the involvement of human subjects. Eligibility for enrollment was open to both male and female participants, with gender not being a consideration in the study’s design. Participants’ sex was determined based on self-reporting. The participants were recruited between September 2021 and March 2023 at our institute. The enrolled participants provided written informed consent. The inclusion criteria were as follows: 1) participants who were diagnosed with malignant tumors, 2) participants who were scheduled to receive immunotherapy based on a decision made by a multidisciplinary team, 3) participants who had no prior immunotherapy, 4) participants who had no regional therapy within 3 months, and 5) participants aged ≥18 years. The exclusion criteria were as follows: 1) participants with a concurrent disease that would impede the treatment regimen and 2) participants who were unable or unwilling to provide written informed consent.

Treatment

Based on initial histological findings and staging, the treatment options were as follows: For checkpoint inhibitors alone or with chemotherapy, patients were treated for 4–6 cycles every 3 weeks until disease progression or intolerable adverse reactions occurred. For CAR-T cell therapy, following leukapheresis and CD19 CAR-T cell (Relma-cel) production, patients were subjected to a low-dose conditioning chemotherapy regimen consisting of fludarabine (40 mg/m² of body surface area per day) and cyclophosphamide (400 mg/m² per day) on days −5, −4, and −3 and then received an intravenous infusion of Relma-cel on day 0.

PET/CT protocols

Baseline ¹⁸F-FDG PET/CT was performed before treatment. Interim ¹⁸F-FDG PET/CT and ⁶⁸Ga-grazytracer PET/CT were performed after 2–3 cycles of immune checkpoint blockade therapy or 1 month post-CD19 CAR-T cell therapy. The interval between the interim ¹⁸F-FDG PET/CT and ⁶⁸Ga-grazytracer PET/CT was within 2 weeks. Interim ¹⁸F-FDG PET/CT and ⁶⁸Ga-grazytracer PET/CT were performed separately. The minimum time between scans was >24 h.

¹⁸F-FDG was synthesized with ¹⁸F and produced using an HM-20 medical cyclotron (Sumitomo Corporation, Japan), with a radiochemical purity of >95%. ⁶⁸Ga-grazytracer was synthesized using a previously described method³⁰. Briefly, 30 nmol of the labeling precursor was dissolved in 300 μL of 0.1 M sodium acetate buffer (pH 5.5) and mixed with 555 MBq of ⁶⁸GaCl₃. This mixture was subjected to a reaction at 99 °C for 10 min and subsequently purified with Sep-Pak C18 cartridges. The final radiotracer was filtered through 0.22-μm filters, and the radiochemical purity was assessed through analytical radio-high-performance liquid chromatography (radio-HPLC). The ¹⁸F-FDG and ⁶⁸Ga-grazytracer doses were 5.55 MBq/kg and 3.70 MBq/kg, respectively. Before the ¹⁸F-FDG PET/CT scan, the participants fasted for ≥6 h to ensure their blood glucose levels were <11.0 mmol/L. No special preparation was required for ⁶⁸Ga-grazytracer PET/CT.

The static ¹⁸F-FDG PET/CT and ⁶⁸Ga-grazytracer PET/CT were performed 1 h and 40 ± 10 min post-injection, respectively. The same scanning parameters were used for the PET/CT. CT was performed using the following parameters: tube voltage, 120 kV; current, 146 mA; and section thickness, 3 mm. PET was acquired in the continuous bed motion mode at 1.5 mm/s from the top of the skull to the upper thigh of each participant. PET data were reconstructed using the ordered subset expectation maximization (OSEM) algorithm with 2 iterations and 11 subsets, using CT data for attenuation correction. Static PET/CT with ⁶⁸Ga-grazytracer and ¹⁸F-FDG was performed using a Biograph mCT Flow 64 scanner (Siemens Healthcare).

For dynamic imaging of ⁶⁸Ga-grazytracer, a low-dose CT (120 kV, 46.13 mAs on average, matrix 512 × 512) was performed to correct attenuation. Subsequently, a dynamic whole-body PET was performed for 40 min post-intravenous ⁶⁸Ga-grazytracer injection through a peripheral vein in the wrist. A total-body PET/CT μEXPLORER scanner (UNITED IMAGING, Shanghai, China) with a maximized PET field of view (FOV) of 194 cm was used. The reconstruction protocols included an OSEM algorithm, 3 iterations, 20 subsets, a 192 × 192 image matrix, 600 mm FOV, 2.886 mm slice thickness, and a 4.5 mm Gaussian post-processing filter. Raw PET data were reconstructed into 95 frames (15 frames × 2 s, 30 frames × 5 s, 30 frames × 14 s, and 20 frames × 90 s). In four patients, a static whole-body PET/CT examination was performed 90 min post-injection for approximately 20 min.

Image analysis

¹⁸F-FDG and ⁶⁸Ga-grazytracer PET/CT studies were reviewed in consensus by ≥2 nuclear medicine physicians (≥10 years of experience) blinded to the clinical data and results of other imaging studies. To decrease bias, all images were re-evaluated by each physician at monthly intervals. The static PET/CT images were co-registered and displayed using a Siemens workstation (Syngo.via VB20, MM Oncology). For quantitative PET/CT analysis, circular regions of interest were manually drawn around the tumor lesions on transaxial slices of the PET images, and the SUV_max was automatically calculated to evaluate tracer uptake in the tumors semi-quantitatively. The SUV_peak was calculated using a 1.2 cm region of interest positioned at the tumor’s hottest site and subsequently normalized to the SUV corrected for lean body mass (SUL_peak) (calculated as SUV_peak × [lean body mass]/[total body mass]).

The dynamic PET/CT images were processed on a uWS-MI R002.20.0.1386284 workstation, where they were meticulously registered and fused with the CT images to ensure precise anatomical alignment. A series of 95 PET frames were combined into a dynamic sequence to facilitate a quantitative analysis of the ⁶⁸Ga-grazytracer uptake kinetics. Nine specific organs (including the brain, heart, lungs, spleen, liver, kidneys, bone marrow, muscle, and skin) were chosen for detailed examination. The volumes of interest (VOIs) were carefully outlined within these organs and tumors on the reconstructed dynamic PET images, with precise adjustments made by referencing the corresponding CT images. Following this process, time-activity curves were derived for each of the specified regions.

For CT analysis, two board-certified radiologists (≥10 years of experience) reviewed the contrast-enhanced CT images and reached an agreement without knowledge of patients’ additional imaging or clinical data. Contrast-enhanced CT was performed at baseline and then every 3 months until disease progression, death, or withdrawal from the study.

Response evaluation

We performed ¹⁸F-FDG PET interpretation according to the 2014 Lugano⁴⁴ (for lymphomas) and PERCIST criteria¹¹ (for solid tumors). Efficacy was categorized as CMR, PMR, stable metabolic disease (SMD), or progressive metabolic disease (PMD). To facilitate comparison between the various criteria, ¹⁸F-FDG PET/CT findings were classified as either metabolic (CMR + PMR) or metabolic non-response (SMD + PMD) after primary interpretation. The interval between the interim contrast-enhanced CT and PET/CT was within 2 weeks. CT results were interpreted according to the 2014 Lugano criteria (for lymphomas) and RECIST 1.1³ (for solid tumors). Efficacy was categorized into CR, PR, SD, and PD. Based on the treatment outcomes, the patients were further classified into response (CR + PR) and non-response groups (SD + PD). The overall response was determined based on CT results 6 months post-treatment and evaluated using the RECIST 1.1 or Lugano criteria.

Immunohistochemistry

Histological diagnosis was performed by two pathologists blinded to the clinical and PET findings. Patient tumor sections from biopsy samples before treatment were stained with the PD-L1 pharmDx kit (clone: 22C3; Agilent Technologies, Santa Clara, CA, USA) and CD8 antibody (1:1000; catalog#12068; Servicebio, Wuhan, China), respectively. For granzyme B staining, tumor sections from surgical samples were stained with an anti-human granzyme B antibody (1:200; catalog#17215; Cell Signaling Technology, Beverly, MA, USA). After incubation with horseradish peroxidase-conjugated secondary antibodies, all sections were visualized by incubating with a diaminobenzidine substrate. Immunohistochemical images were obtained using Slide viewer (version 2.7), and the data were analyzed with Image J software (version 1.52p).

Tumor sections were assessed, with positive PD-L1 expression being classified as a tumor proportion score of 1% or above. CD8⁺ T cell infiltration was characterized as either a “desert” or “non-desert” phenotype⁴⁵, as defined by the criteria outlined as follows. Tumor phenotype was designated as “desert” if the frequency of CD8⁺ T cells was low and specifically fewer than 10 CD8⁺ T cells per high-power field (200× magnification); for larger specimens, this was determined by averaging counts from 10 randomly selected high-power fields. Conversely, tumors were labeled as “non-desert” when the number of CD8⁺ T cells in the field of view surpassed this threshold of 10.

The granzyme B positivity rate was determined by quantifying the percentage of stained cells positive for granzyme B out of the total number of cells counted. This involved assessing a minimum of 10 randomly chosen unit areas per sample. In each unit area, the number of cells positive for granzyme B was recorded. The positivity rate was then computed by summing the counts of granzyme B-positive cells across all unit areas and dividing by the cumulative total of cells counted within those fields.

Statistical analysis

Statistical analyses were conducted using GraphPad Prism 9 (GraphPad Software, San Diego, CA, USA) and R software (version 4.3.3; available at www.r-project.org). Descriptive statistics were employed to summarize the demographic data of patients. Non-linear approaches were utilized to compare the biodistribution of ⁶⁸Ga-grazytracer among various healthy organs. Specifically, the General Additive Mixed Model (GAMM) was applied to examine variations in ⁶⁸Ga-grazytracer uptake in different organs over time. Fixed effects in the model comprised time and organ (as categorical variables including brain, heart, kidney, liver, lung, muscle, skin, and spleen), while a random intercept was included for each patient to account for the within-subject variability due to repeated measurements. Furthermore, a quadratic (second-degree polynomial) model was used to illustrate the time-dependent biodistribution curves.

Pearson correlation analysis was used to assess the association between the uptake of ⁶⁸Ga-grazytracer in tumors and the levels of granzyme B, as measured by IHC staining of the corresponding tumor tissue sections. The relationships between PD-L1 IHC findings and the overall response, as well as between different imaging assessment criteria and the overall response, were analyzed utilizing Chi-square test and Fisher’s exact test. ROC curves were generated to assess the specificity and sensitivity of ⁶⁸Ga-grazytracer PET in predicting clinical outcomes, compared to CT and ¹⁸F-FDG PET; P values were determined using the Wilson/Brown method.

In the analysis of PFS, the time to event was defined as the duration from the start of the initial treatment (time zero) to the occurrence of tumor progression or death from any cause. The data of patients who did not experience disease progression by the final tumor assessment were censored. The log-rank test was utilized to assess the statistical significance of differences in PFS between groups. Additionally, the Cox regression model was used to estimate the hazard ratios. To investigate potential violations of the proportional hazards assumption, the Schoenfeld residual test was conducted.

Other comparisons were performed using the Mann–Whitney U test to calculate P values. All reported P values are two-tailed, with statistical significance established at P < 0.05.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.