Immune checkpoint inhibitors including anti-programmed death-1 (anti-PD-1) antibody play a central role in the treatment of advanced non-small cell lung cancer (NSCLC). Although a significant number of patients with cancer benefit from anti-PD-1 therapy, many fail to have clinical responses.1 To date, the most successful biomarker associated with response to anti-PD-1 therapy is tumor cell expression of programmed death-ligand 1 (PD-L1).2 3 However, more than half of NSCLC patients with high PD-L1 expression (≥50%) do not respond to anti-PD-1 therapy, while 9%–17% of those without PD-L1 expression (<1%) do respond.4–6

The therapeutic response to immunotherapy is generally assessed by morphological changes using CT. However, it is difficult to assess the response to immunotherapy at an early phase using CT because of two immunotherapy-specific responses. First, a phenomenon known as pseudoprogression has been described in patients with various tumors who receive immunotherapy and experience an objective response after initial tumor progression.7 8 Second, a paradoxical acceleration in tumor growth early after immunotherapy initiation known as hyperprogression,9 which is correlated with worse prognosis, has been reported. Therefore, when tumors grow early after anti-PD-1 therapy initiation, it is difficult to determine whether the growth is true or not. For these reasons, some radiological criteria for immunotherapy have been recently developed to assess tumor responses correctly.10–12 According to these criteria, a confirmatory CT at 4–8 weeks after the first findings of progression by CT is required to distinguish progressive disease (PD) from pseudoprogression. Thus, it is impossible to predict the early response to immunotherapy using CT. Therefore, there is an urgent need to establish more accurate biomarkers for the response to immunotherapy, which can help to curtail ineffective and potentially toxic therapies.

[¹⁸F ]-Fluoro-2-deoxy-D-glucose (¹⁸F-FDG) positron emission tomography (PET) has recently been applied to assess the response to chemotherapy in various tumor types.13 One important problem in the use of ¹⁸F-FDG PET to evaluate the response to immunotherapy is ¹⁸F-FDG accumulation in activated glucose-consuming inflammatory cells in and around tumors during immunotherapy. In fact, it has recently been reported that resected lung cancer lesions with major pathological responses to anti-PD-1 neoadjuvant therapy show increased CD8+ T cell and macrophage infiltration into the tumor.14 As a result, in our previous study, 33.3% of responders showed an increase in ¹⁸F-FDG uptake 2 weeks after anti-PD-1 therapy.15 Thus, there is an increasing clinical need to use other PET tracers to avoid this disadvantage.

3′-Deoxy-3′-[¹⁸F ]-fluorothymidine (¹⁸F-FLT) PET has been proposed as an imaging method that can evaluate cell proliferation indirectly,16 since a significant correlation has been found between ¹⁸F-FLT uptake and Ki-67 in various tumors.17 ¹⁸F-FLT PET has been described as superior to ¹⁸F-FDG PET for the quantitative assessment of tumor cell proliferation.18 Recently, ¹⁸F-FLT PET imaging has been shown to be useful in predicting treatment response to chemotherapy, chemoradiotherapy, and molecularly targeted drugs.19 20 However, to date, there is no report applying ¹⁸F-FLT PET imaging to evaluate early responses to immunotherapy for NSCLC. The aim of this study was to evaluate whether changes in tumor ¹⁸F-FLT accumulation before and after the initiation of anti-PD-1 therapy could accurately predict tumor response as assessed by immune-related Response Evaluation Criteria in Solid Tumor (irRECIST) criteria21 and patient survival.

This is the first prospective study to investigate the usefulness of ¹⁸F-FLT PET as an early predictive biomarker for the response to anti-PD-1 therapy in patients with advanced NSCLC. To our knowledge, there are three reports regarding immunotherapy response assessment with ¹⁸F-FLT PET in patients with melanoma27–29 and one report in those with prostate cancer.30 However, the sample sizes of these studies were small (n=5–17), and thus, the predictive capability and optimal timing of repeated ¹⁸F-FLT PET in these situations are not well elucidated. The present study showed that the change in ¹⁸F-FLT accumulation between baseline and 2 weeks of treatment (ΔTLP0-2) had a moderate predictive value for objective responses as assessed by irRECIST, and the changes between baseline and 6 weeks after treatment initiation (ΔSUV_max0-6, ΔPTV0-6, and ΔTLP0-6) were able to predict responses more precisely as compared with ΔTLP0-2. Response assessments using ¹⁸F-FLT PET parameters may be able to assess treatment efficacy earlier than CT-based irRECIST. Furthermore, both ΔTLP0-2 and ΔTLP0-6 were significant prognostic factors for PFS, whereas known biomarkers such as tumor PD-L1 expression, ECOG-PS, and age were not.

Several previous studies have demonstrated the efficacy of ¹⁸F-FDG PET for predicting the response to immunotherapy in patients with advanced NSCLC. In many of these studies, serial PET scans were planned at approximately 2 months after treatment initiation.31 32 However, since some phase III trials using anti-PD-1 antibody in patients with NSCLC have shown that the median time to response is 8–9 weeks,4–6 the clinical benefits of the first response evaluation using PET scans at 2 months may be limited. In a recent study with a short-term evaluation, Kaira et al33 showed that the metabolic response by ¹⁸F-FDG uptake at 1 month after nivolumab treatment successfully predicted objective responses and survival. Furthermore, we recently reported the relationship between the change of ¹⁸F-FDG uptake at 2 weeks after nivolumab treatment and objective responses in patients with previously treated NSCLC.15 However, in this study, three of the nine responders exhibited an increased ¹⁸F-FDG uptake (>20%) between baseline and at 2 weeks, and thus ¹⁸F-FDG accumulation in tumor-infiltrating immune cells may cause an inflammatory metabolic pseudoprogression in early-phase PET imaging after anti-PD-1 therapy. Since ¹⁸F-FLT accumulation is considered to be a marker of cell proliferation, we expected that the predictive value of the changes in ¹⁸F-FLT PET parameters after 2 weeks would be superior to that of ¹⁸F-FDG PET parameters. However, the predictive accuracy of a ¹⁸F-FLT PET parameter (ΔTLP0-2) to predict PD as assessed by irRECIST in the current study was similar to that of the ¹⁸F-FDG PET parameter in our previous study (76.0% vs 76.0%, respectively),15 although the patient populations of those studies were not the same. However, the changes in tumor ¹⁸F-FLT uptake at 6 weeks after therapy showed a high accuracy of 90.9% for the prediction of PD and a significant correlation with PFS. Similarly, our previous study found that patients without relapse-associated findings on ¹⁸F-FDG PET at 8 weeks had significantly longer PFS than patients with PMD.15 Although a simple comparison of these two studies is impossible, serial ¹⁸F-FLT PET might be able to predict tumor response as quickly as or earlier than ¹⁸F-FDG PET. However, needless to say, ¹⁸F-FDG PET is more widely used than ¹⁸F-FLT PET, and there is more evidence for its effectiveness for tumor response assessment.

Among the response patterns we observed, three of eight patients who achieved PR showed increased TLP from baseline to 2 weeks. Of note, this discrepancy of response was not observed in 6-week ¹⁸F-FLT PET imaging. To our knowledge, this is the first report demonstrating the phenomenon of ‘proliferative’ pseudoprogression with ¹⁸F-FLT PET imaging. In addition to proliferative tumors, physiological ¹⁸F-FLT uptake has been observed in the bone marrow because of the high proliferative activity of hematopoietic cells in the marrow.23 34 Furthermore, ¹⁸F-FLT PET can also detect proliferative changes in spleens and lymph nodes after immunotherapy.27–30 Therefore, ¹⁸F-FLT uptake may be a reliable marker to assess the proliferative activity of immune cells. A recent study described the early pathological changes after neoadjuvant nivolumab therapy in patients with NSCLC.14 In that report, primary tumors with a major pathological response showed large numbers of infiltrating CD8+ T cells and macrophages. Meanwhile, an increase in the proliferation of CD8+ T cells in the blood within 4 weeks of anti-PD-1 therapy was observed in patients with advanced NSCLC, and it suggested an association with clinical benefit for that therapy.35 Thus, PD-1 blockade may induce the proliferation of CD8+ T cells in responding tumors and cause proliferative pseudoprogression in ¹⁸F-FLT PET imaging.

In the present study, one of five patients with dissociated response had an incorrect prediction for early response to immunotherapy on repeated ¹⁸F-FLT PET imaging. It has been reported that 7.7%–17.7% of NSCLC patients who are treated with immunotherapy experience dissociated responses in CT imaging.36 37 Some recent studies have indicated possible biological explanations for dissociated responses. First, there are specific differences in the immune tumor microenvironment at different metastatic sites,38 and these differences, including tumor-infiltrating lymphocyte status, can influence the response to immunotherapy.39 Second, since recent evidence suggests that multiple cytokines such as interferon-γ, IL-1α, and IL-27 induce or enhance PD-L1 expression in tumor cells,40 and the expression of these cytokines is influenced by the tumor microenvironment, PD-L1 expression can be heterogeneous among metastatic sites.36 37 Although dissociated responses in ¹⁸F-FLT PET imaging may lead to incorrect predictions of early response to immunotherapy, it can be advantageous to examine the changes in the proliferation of each lesion. Recently, new PET tracers, such as ⁸⁹Zr-C4 and ¹⁸F-BMS-986192, that visualize tumor PD-L1 expression have been developed and show good correlation with tumor PD-L1 expression as measured by IHC and responses to immunotherapy.41 42 Since these PET images can reveal the heterogeneity of PD-L1 expression between different tumor lesions, they may predict dissociated responses prior to treatment initiation.

We found that ΔTLP was a better predictor for efficacy of anti-PD-1 therapy than ΔSUV_max, which is in line with some previous reports using PET imaging for the response assessment of immunotherapy.13 15 31 Since it has been reported that PD-L1 expression is highly heterogeneous within tumors,36 the heterogeneity of tumor PD-L1 expression may reflect heterogeneous infiltration of inflammatory cells and cause a dissociated result between SUV_max and TLP in response to anti-PD-1 therapy. Thus, volumetric analysis (PTV or TLP) of PET imaging may be a more appropriate method for the response assessment to immunotherapy as compared with SUV_max.

This study has some limitations. ¹⁸F-FLT PET is limited when assessing liver and bone metastases because of the high background uptake in bone marrow and liver. Another limitation is the small sample size, so that the reliability of the cut-off value of this study is not certain. Furthermore, as described in the Materials and methods section, we examined various thresholds and decided that an SUV of 2.0 was most suitable for volumetric analysis of ¹⁸F-FLT PET in patients with NSCLC, though full validation of this SUV threshold is needed in the future. However, the present initial observation is valuable for future studies to establish appropriate timing of follow-up ¹⁸F-FLT PET scanning, a cut-off value for response assessment, and an SUV threshold for volumetric analysis. Further studies with larger samples are required to examine the efficacy of ¹⁸F-FLT PET in predicting the early response to immunotherapy.

The results of this study suggest that ¹⁸F-FLT PET is a useful tool for predicting early response to anti-PD-1 therapy in patients with NSCLC. The changes in ¹⁸F-FLT uptake as early as 2 weeks after treatment initiation had a moderate predictive value for subsequent treatment efficacy, although proliferative pseudoprogression was observed in 37.5% of responders. However, the changes between baseline and 6 weeks after treatment initiation were robust biomarkers of treatment efficacy. Further studies with larger sample sizes would be beneficial to validate this conclusion.