T‐cell response to phytohemagglutinin in the interferon‐γ release assay as a potential biomarker for the response to immune checkpoint inhibitors in patients with non‐small cell lung cancer

Abstract Background Immune checkpoint inhibitors are a standard treatment for advanced lung cancer, although it remains important to identify biomarkers that can accurately predict treatment response. Immune checkpoint inhibitors enhance the antitumor T‐cell response, and interferon‐γ plays an important role in this process. Therefore, this study evaluated whether the number of interferon‐γ‐releasing peripheral T cells after phytohemagglutinin stimulation in the interferon‐γ release assay might act as a biomarker for the response of non‐small cell lung cancer to immune checkpoint inhibitor treatment. Methods Data were retrospectively collected regarding 74 patients with non‐small cell lung cancer who had received immune checkpoint inhibitors. Pretreatment screening tests had been performed using the T‐SPOT.TB assay, which quantifies the number of interferon‐γ‐releasing T cells (as immunospots) in response to phytohemagglutinin and tuberculosis‐specific antigen stimulation. Clinical factors and the number of spots in the T‐SPOT fields were evaluated for associations with patient outcomes. The median number of spots was used to categorize patients as having high or low values, and the two groups were compared. Results Relative to patients with a low ratio, patients with a high ratio of phytohemagglutinin/tuberculosis‐specific antigen spots (i.e. more responsive T cells) had significantly better progression‐free survival after immune checkpoint inhibitor treatment. When we only considered patients with negative T‐SPOT results, a high number of phytohemagglutinin‐stimulated spots corresponded to significantly longer progression‐free survival. Conclusion The T‐SPOT.TB assay can be used to quantify the number of immunospots in response to antigen stimulation, which may predict the response to immune checkpoint inhibitors in patients with non‐small cell lung cancer.


BACKGROUND
Lung cancer is one of the most common cancers and is associated with high rates of morbidity and mortality. However, the introduction of immune checkpoint inhibitors (ICIs) has helped improve the prognosis of patients with lung cancer. [1][2][3][4][5][6] In addition to ICI monotherapy, combined treatment using an ICI and cytotoxic anticancer drugs has been approved as standard therapy for non-small cell lung cancer based on results from clinical trials. 7 Tumor cell expression of programmed death-ligand 1 (PD-L1) is a commonly used biomarker for predicting the response to ICI therapy. However, this biomarker has limited utility, as some patients with high PD-L1 expression experience a poor response to ICI treatment, and it can be difficult in some cases to obtain a tumor specimen and evaluate PD-L1 expression. 8 The effects of ICI treatment are mediated by T cells, and T-cell dysfunction can lead to treatment resistance. 9 Interferon-γ (IFN-γ) is released from T cells as part of their normal activation, and IFN-γ concentrations are reportedly related to ICI treatment response. 10 Furthermore, disruption of the IFN-γ receptor signaling pathway can lead to resistance to anti-PD-L1 therapy. 11 An IFN-γ releasing assay (IGRA) can be used to identify Mycobacterium tuberculosis based on T-cell production of IFN-γ in response to stimulation with tuberculosis-specific antigens (TBAgs). As a positive control, the T-cell response is quantified after nonspecific stimulation using phytohemagglutinin (PHA). Thus, the ability of T cells to produce IFN-γ in response to PHA stimulation may reflect the immune status of individual patients. 12 The T-SPOT.TB test is a type of IGRA that evaluates T-cell production of IFN-γ using an enzyme-linked immunospot count after antigen stimulation. Therefore, the present study evaluated whether the T-SPOT test could be used to quantify IFN-γ production by peripheral T cells and thus predict the response to ICI treatment in patients with non-small cell lung cancer.

Patients
This retrospective study evaluated data from adult patients with non-small cell lung cancer who received ICI treatment at the Yokohama City University Medical Center, Minami Kyosai Hospital, and Yokohama City University Hospital between January 2016 and August 2019. The inclusion criteria were age >18 years, a diagnosis of stage IV or postoperative recurrence non-small cell lung cancer, ICI treatment (nivolumab, pembrolizumab, or atezolizumab), and available IGRA test results, which had been used to screen for tuberculosis infection before starting anticancer chemotherapy in any line. The exclusion criteria were a diagnosis of small cell lung cancer and clear signs of infectious disease at the start of ICI treatment. Only the first ICI treatment was considered in patients who received multiple ICI treatments. Data regarding progression-free survival (PFS), overall survival (OS), patient characteristics, pathological type, PD-L1 tumor proportion score (TPS), the categorical T-SPOT test result, and the number of immunospots in each T-SPOT field were collected. The study's retrospective protocol was approved by our institutional review board (B191200043), which waived the requirement for informed consent.

Treatment regimens
Patients were considered eligible if they had received ICIs as monotherapy or in combination with chemotherapy. Nivolumab monotherapy was administered intravenously at a dose of 3 mg/kg every 2 weeks. Pembrolizumab monotherapy was administered intravenously at a dose of 200 mg/kg every 3 weeks. Atezolizumab monotherapy was administered intravenously at a dose of 1200 mg/kg every 3 weeks. The combination regimens involved pembrolizumab with cisplatin/ carboplatin and pemetrexed/paclitaxel.

T-SPOT.TB assay
The T-SPOT tests were performed according to the manufacturer's protocol. Pretreatment peripheral blood samples were collected into heparinized tubes to isolate peripheral blood mononuclear cells (PBMCs). The isolated PBMCs were incubated in a microplate well on which IFN-γ antibody was immobilized with or without TBAgs (ESAT-6 or CFP-10) or PHA for nonspecific stimulation as a positive control. About 2.5 Â 10 5 PBMCs were added to each well, and the incubation conditions were 37 C, and 5% CO 2 for 16-20 h. After the incubation, the secondary antibody was added and allowed to react at 4 C for 1 h. The number of spots was counted in the T-SPOT test, and the results were graded according to the categories described in the manual. The median number of T-SPOT immunospots was used to categorize results as having high and low numbers of immunospots.

Statistical analysis
Baseline patient characteristics were extracted from their medical records. The PFS interval was calculated from the start of ICI treatment to the first instance of disease progression (defined according to version 1.1 of the Response Evaluation Criteria in Solid Tumors) or death because of any cause. The OS interval was calculated from the start of ICI treatment to death or the last visit date. Curves for PFS and OS were compared using the Kaplan-Meier method and the log-rank test. A Cox proportional hazards model was used for the multivariable analysis, and differences were considered statistically significant at p values <0.05. All analyses were performed using JMP Pro 15 software (SAS Institute Inc.).

Baseline characteristics
A total of 392 lung cancer patients were assessed for eligibility ( Figure 1). However, we excluded 41 patients with small cell lung cancer, 186 patients who did not receive ICI treatment, and 91 patients who did not undergo T-SPOT testing. Thus, the study included data from 74 patients who received ICI treatment and had T-SPOT test results.
The baseline characteristics are summarized in Table 1. The median age was 67 years (range 34-83 years) and 60 patients were men. Most patients had a good performance status (PS), with 38 patients having a PS of 0, 26 patients having a PS of 1, nine patients having a PS of 2, and one patient having a PS of 3. When ICI therapy was performed during the lung cancer treatment, 24 patients were treated in the first line, 26 patients in the second line, and 24 patients after third-line treatments. The treatment regimens of ICI involved nivolumab monotherapy (34 patients), pembrolizumab monotherapy (30 patients), atezolizumab monotherapy (five patients), and pembrolizumab in combination with platinum-based chemotherapy (five patients). The PD-L1 TPS values were >50% for 23

Correlation between spot counts and response to ICI treatment
We evaluated the patients' characteristics and spot counts for relationships with PFS. Patients were classified into two groups using the median number of spots in Abbreviations: CFP-10, a tuberculosis-specific antigen; ESAT-6, a tuberculosis-specific antigen; PD-L1, programmed death-ligand 1; PHA, phytohemagglutinin; TBAg, tuberculosis-specific-antigen; TPS, tumor proportion score. response to PHA divided by the median number of spots in response to the TBAgs. This division was performed to rectify an immune response to TBAgs. When there were no spots in response to TBAgs, we used the number of PHA spots as the denominator. Figure 2(a) shows the PFS curves when we categorized the PHA/ESAT-6 spot counts as high or low (median PFS 323 days vs. 51 days; p = 0.006). Figure 2(b) shows the PFS curves when we categorized the PHA/CFP-10 spot counts as high or low (median PFS 285 days vs. 63 days; p = 0.036). The patients in the PHA/TBAg high and low groups had generally similar characteristics, with the exception of unknown PD-L1 status (Table 3). Given the larger number of patients with unknown PD-L1 status in the PHA high group, multivariable Cox proportional hazards analysis was performed for PS, PD-L1 TPS, and age (known predictors of response to ICI therapy and chemotherapy). The results revealed that better PFS was associated with better PS, higher PHA/ESAT-6 spot count, and higher PD-L1 TPS (Table 4). Similar to PFS, the analysis of the relationship between OS and PHA/TBAg spots, an increase in the number of PHA/ TBAg spots were associated with a longer OS (Supporting Information Figure S1, and Supporting Information Tables S1 and S2).

The PHA-induced spot count independently predicts PFS among patients with negative T-SPOT results
Given that T-cell response to TBAgs might be involved in their response to mitogen activation, we excluded patients with positive, intermediate, and indeterminate T-SPOT results and only considered patients with negative T-SPOT results. Patients in the PHA high and low groups had similar characteristics, except for PD-L1 status ( Table 5). The PFS outcomes were compared between the PHA high and low groups (Figure 3), which revealed that the PHA high group had significantly longer median PFS (886 days vs. 73.5 days; p = 0.019). Multivariate analysis was performed because the PD-L1 status differed between the two groups (Table 6), and the results showed that better PFS was associated with better PS and a high PHA spot count. Similarly for OS, a high number of PHA spots was associated with better OS (Supporting Information Figure S2 and Table S3).

DISCUSSION
The increasing use of ICI treatment for lung cancer patients highlights the need for biomarkers that accurately predict treatment response, which can help minimize ineffective treatment and prevent adverse effects that are often severe. This study evaluated whether the number of IFN-γ-releasing cells based on T-SPOT counts could predict the response to ICI treatment, and we observed that the PHA/TBAg spot count (number of IFN-γ-releasing cells in response to PHA/TBAg stimulation) was significantly associated with PFS after starting ICI treatment for non-small cell lung cancer ( Figure 2 and Table 4). Moreover, among patients with negative T-SPOT results, a high PHA spot count was associated with significantly better PFS (vs. a low PHA spot count) ( Figure 3 and Table 6). These results showed a similar trend for OS (Supporting Information Figure S1, S2 and Table  S1-S3). Interferon-γ is an important factor that can exert broad pro-or antitumor effects on the immune response. 13 For example, IFN-γ upregulates the MHC class I beta-2-microglobulin in tumor cells 14,15 and IFN-γ also improves the antitumor immune response of NK cells and cytotoxic T-lymphocytes. [16][17][18] Nevertheless, another group has reported that IFN-γ can downregulate the antitumor immune response by promoting regulatory T-cell (Treg) function and suppressing effector T-cell function via enhancement of indoleamine 2,3-dioxygenase secretion from cancer cells and antigen-presenting cells. 19 Cho et al. have also reported that IFN-γ activates the nonclassical MHC class Ia genes and causes evasion of cytotoxic lymphocytes in melanoma. 20 In the context of the PD-1/PD-L1 axis, IFN-γ induces PD-L1 expression on both cancer cells and local immune cells. 21,22 Thus, IFN-γ appears to play two opposing but important roles in regulating the antitumor immune response.
Our results indicate that the number of IFN-γ-releasing peripheral T cells, based on T-SPOT counts, might be useful for predicting the response to ICI treatment among patients with non-small cell lung cancer. In this context, patients who have T cells that produce large amounts of IFN-γ might be more likely to experience a favorable response to ICI treatment. Huang et al. evaluated 84 patients who received cytotoxic therapy for lung cancer and reported that higher IFN-γ production after PHA stimulation was associated with better outcomes, 23 although we are only aware of two small studies that have evaluated the relationship between IFN-γ production and response to ICI treatment.
Karachaliou et al. evaluated serum mRNA expression of IFN-γ before nivolumab treatment, which revealed that high pretreatment expression was associated with significantly longer median PFS (high group [n = 11] 5.1 months vs. low group [n = 4] 2.0 months; p = 0.0124). 24 Kanai et al. used the QuantiFERON-TB Gold Plus (QFT-Plus, Qiagen) assay, which evaluates IFN-γ levels using ELISA, 10 and reported that a significantly longer PFS was observed for the high IFN-γ group (n = 24) relative to the low IFN-γ group (n = 5). Our data, which were derived from a larger sample of patients, strongly support the relationship between high IFN-γ production and prolonged PFS in this setting. Our study was based on the T-SPOT test, which evaluates IFN-γ production based on immunospots after PBMCs are exposed to different antigens. The T-SPOT results are adjusted for the peripheral blood lymphocyte count, which can help reduce the influence of varying lymphocyte counts between patients. 25 There is evidence that ICI treatment is associated with a risk of developing tuberculosis, 26 positive T-SPOT results in 6.8% of our patients who received ICI treatment for non-small cell lung cancer (Table 2). However, this proportion is comparable to the proportion of positive results that we previously reported among all outpatients at our institution (8.4%). 29 Although none of our patients developed active tuberculosis during or after ICI treatment, ICI is still considered a risk factor for the development and aggravation of tuberculosis. 30 Therefore, it may be prudent to consider prophylactic treatment for patients with positive T-SPOT results.
In this study, the timing of the T-spot test was not limited to before the first line. A report on T-SPOT tests 29 has confirmed the reproducibility of its results and the timing of blood collection might not be a problem. To examine the details, a unified study of the timing of blood collection should be considered.
This study has several limitations that should be considered. First, we only considered a small sample of patients from a few Japanese institutions. Second, the retrospective study design is prone to bias and incomplete data. Third, the T-SPOT results reflect the reaction of PBMCs, and further studies are needed to confirm the relationship between the tumor microenvironment and peripheral blood findings.
In conclusion, the present study revealed that based on the T-SPOT test results, the number of IFN-γ-releasing peripheral T cells after PHA stimulation might be useful for predicting the response to ICI treatment among patients with non-small cell lung cancer. Thus, a larger prospective study is warranted to confirm whether the response of peripheral blood-derived T cells to PHA is a useful biomarker for predicting the response to ICI treatment.