A Validation Study on IDO Immune Biomarkers for Survival Prediction in Non–Small Cell Lung Cancer: Radiation Dose Fractionation Effect in Early-Stage Disease

Lung cancer is the leading cause of cancer-related death in the United States and worldwide, with non–small cell lung cancer (NSCLC) accounting for 80% to 85% of cases (1). Definitive thoracic radiotherapy (RT) is a standard treatment option in medically inoperable early-stage NSCLC patients and is associated with low treatment-related morbidity (2). However, treatment outcomes for early-stage NSCLC after RT remain suboptimal. RT has recently been reported to activate the host immune system and induce an abscopal effect (3–5). In addition, in a subgroup analysis of a phase I trial in patients with metastatic NSCLC treated with pembrolizumab, patients who received RT prior to immunotherapy had significantly better OS than those without prior RT (6). Therefore, in addition to direct tumor killing via DNA damage, RT may modulate the immune tumor microenvironment (TME). On the other hand, RT also causes immunosuppression through direct toxicity to primary and secondary lymphoid organs as well as circulating immune cells (7). RT fractionation schedule is directly related to the risk and severity of RT-induced immunosuppression, as has been shown by studies comparing the risk of severe radiation-induced lymphopenia following hypofractionated versus conventionally fractionated RT (8–10). However, the relationship between RT fractionation scheme and other biomarkers of immune function is not well understood.

Indoleamine-2,3-dioxygenase (IDO) is a heme-containing endogenous enzyme that catalyzes the initial and rate-limiting steps in the kynurenine pathway, which converts tryptophan into kynurenines (11). IDO is expressed in many cancer types and has been established as an immune suppressor in cancer initiation and progression (12). Furthermore, high IDO expression has consistently been found to be associated with poor clinical outcomes in solid tumors (13). We recently demonstrated that chemoradiotherapy significantly reduced IDO activity during the initial phase of treatment (2–4 weeks during a 6-week RT course) in 110 stage III NSCLC patients. Moreover, baseline IDO activity and changes in IDO levels after RT significantly correlated with survival and tumor progression (14). However, it is not clear how to optimize clinical RT dose/fractionation regimens to maximize RT-induced immunostimulation and/or minimize RT-related immunosuppression. Furthermore, the role of IDO immune status as a biomarker for treatment outcome in early-stage NSCLC treated with definitive thoracic RT remains unknown.

In the present study, we examined the effect of RT dose fractionation on IDO-mediated immune activity in patients with stage I-II NSCLC. We also examined differences in IDO dynamics during RT in patients with early-stage NSCLC treated with different RT fractionation regimens. In addition, to validate previous findings suggesting that IDO is a biomarker for survival in stage III NSCLC, we examined the association between circulating levels of IDO with overall survival (OS), progression-free survival (PFS), and local/regional/distant failure rates in patients with early-stage NSCLC treated with definitive RT.

From 2004 to 2015, patients with stage I–II inoperable/unresectable NSCLC who enrolled in Institutional Review Board-approved prospective protocols were eligible. The protocols were conducted according to requirements of the Belmont Report and U.S. Common Rules. All subjects signed informed written consent before enrollment. All patients received definitive thoracic RT, which was given using either 3DCRT or SBRT as previously described (15, 16). Briefly, the RT dose fractionation regimens were typically 60-85.5 Gy in 2-Gy fractions for 3DCRT or 50, 55, or 60 Gy in 10-, 11-, or 20-Gy fractions for SBRT. Details of these prospective trials are summarized in Supplementary Table S1. Biological equivalent dose (BED) was calculated for each patient using an alpha/beta of 10 to compare tumor prescription doses for different RT fractionations.

Patient follow-up, blood sample processing and measurements of IDO biomarkers were as previously described (14). Patients were examined weekly during the course of RT, followed up approximately every 3 months during the first year, every 6 months during the second year, and then annually thereafter. Clinical and follow-up data, including age, gender, smoking history, histology, clinical stage, tumor volume, Karnofsky performance score (KPS), BED, and chemotherapy, were recorded prospectively for up to 2 years after RT completion. Red top blood collection tubes without anticoagulants were used for serum. Blood samples of patients were collected at up to three time points: before RT (pre-RT), during RT (generally at 2 week for 3DCRT and at fraction number 2/3 or 3/5 SBRT), and post-RT (generally at the time first follow-up, 3 months after RT completion).The samples were stored in a −80°C freezer immediately after collection until testing.

Serum tryptophan and kynurenine concentrations were measured using a high-performance liquid chromatography system (Shimadzu LC20) as previously described, with minor modifications (17, 18). First, 25-μL serum samples were diluted with equal volumes of 30-mmol/L NaAc pH4.0 and deproteinated with perchloric acid. Kynurenine was detected on an UV channel at 360 nm, and tryptophan was detected on a fluorescence channel at 285-nm excitation and 365-nm emission. Samples were analyzed using Lab Solution software (Shimadzu). Samples were quantified with external standards, and at least one quality control sample was randomly inserted into every plate for reference. For quality control purposes, double-blinding duplicate/triplicate testing verified the assay reproducibility to be over 95%.

OS and PFS were the primary endpoints, and local/regional/distant failure rates were the secondary endpoints. OS was calculated from RT start to the date of death of any cause. PFS was defined from start of RT to the date of any progression or death. Patients were censored at last follow-up if progression or death had not occurred. Progression was defined as the first failure type of local/regional/distant progression or death. Non-cancer progression death was defined as death without any evidence of tumor progression at local, regional, or distant sites. The Kaplan–Meier log-rank test was used to compare between-group survival differences. Clinical variables with P < 0.05 by univariate analysis were selected as co-variates for adjustment in the multivariate analysis. A multivariate Cox proportional hazards model was used to estimate hazard ratios (HR) with 95% confidence interval (95% CI). Levels of IDO-checkpoint associated molecules at each of the 3 time points were compared using repeated ANOVA (equivalent to paired t test for 2 groups). Two-tailed tests were performed for statistical significance at a P < 0.05 level. All statistical analyses were performed using IBM SPSS Statistics 22.0 (IBM, Inc.). Scatterplot figures of IDO dynamics were generated using GraphPad Prism (version 5.01).

Of 58 staged I-II NSCLC patients enrolled, 56 patients with adequate serum samples available for IDO testing formed the primary study population; there were 51 white patients (91%) and 14 (25%) female patients. Median age was 68 years (range, 46–85). Twenty-eight patients received definitive-dose (≥ 60 Gy) 3DCRT, 5 (18%) of whom received concurrent platinum-based chemotherapy. Twenty-eight patients underwent SBRT (10-20 Gy per fraction); only one SBRT patient had prior chemotherapy history. Table 1 details patient demographic and clinical features in the 3DCRT and SBRT groups. There was no difference in distributions of age, gender, smoking status, histology type, clinical stage and KPS between the two groups. However, SBRT patients had smaller tumor volume, higher BED and were less likely to receive chemotherapy compared with 3DCRT patients (all P < 0.05).

Median follow-up time was 49 (95% CI, 31–66), 73 (95% CI, 52–94), and 34 (95% CI, 23–45) months for all patients, 3DCRT patients and SBRT patients, respectively. Median OS and PFS were 30 (95% CI, 26–34) and 14 (95% CI, 11–18) months, 30 (95% CI, 25–35) and 14 (95% CI, 10–18) months, and 32 (95% CI, 19–44) and 20 (95% CI, 12–27) months for all patients, 3DCRT patients and SBRT patients, respectively. Univariate analysis showed that patients with smaller tumor volume had better OS (P = 0.017), whereas female patients had better PFS (P = 0.023). Tumor volume and gender were thus selected as clinical co-variates for further multivariate analysis of IDO checkpoint-associated parameters. Chemotherapy and RT fractionation were not significantly associated with OS or PFS in this small series (Table 2, Fig. 1A and B). However, compared with 3DCRT patients, local/regional and distant failure rates were significantly lower in SBRT patients (P = 0.002; Fig. 1C).

Baseline levels of kynurenine or K:T ratio were not significantly associated with OS or PFS in this series. Post-RT kynurenine levels were significantly correlated with OS (HR, 1.25; 95%CI, 1.01–1.55; P = 0.044) but not PFS (HR, 1.11; 95% CI, 0.90-1.36; P = 0.334; Fig. 2A1, Table 3) by multivariate analysis, after adjusting for clinically significant factors. Using median as the cut-off value for changes, higher post/pre kynurenine ratios were correlated significantly with worse PFS (HR, 2.76; 95% CI, 1.17–6.52; P = 0.021; Fig. 2B1, Table 3) but not OS (HR, 2.01; 95% CI, 0.82–4.96; P = 0.129). For local/regional/distant failure rates, patients with greater ratios of post/pre kynurenine had significantly higher rates of local/regional or distant failure, compared with patients with lower levels of post/pre kynurenine, who were less likely to experience progression but had significantly higher rates of death unrelated to cancer progression (P = 0.037; Fig. 1D).

To avoid potential confounding effect of radiation to nodal disease, 5 patients with N1 disease in the 3DCRT group were excluded during group comparison analysis because such patients would be technically ineligible for SBRT. Serum kynurenine and K:T ratio changed heterogeneously through 2 time points from pre-RT, to during-RT and post-RT. In 3DCRT patients, mean concentrations of kynurenine (Fig. 3A) and K:T ratio (Fig. 3B) did not change significantly at during-RT (P = 0.165 and P = 0.194), but increased sharply at post-RT (P = 0.006 and P = 0.017). The concentrations of kynurenine and K:T ratio post-RT were significantly higher than that at other time points. In SBRT patients, however, neither serum kynurenine nor the K:T ratio changed significantly at during-RT or post-RT (compared with baseline; all P > 0.05). No statistically significant differences in IDO parameters were observed between 3DCRT and SBRT patients at any of the studied time points (T test). However, 3DCRT patients had a significantly larger increase in the absolute kynurenine levels following RT than SBRT patients, with a median increase of 0.721 ng/mL in 3DCRT patients vs. 0.115 ng/mL in the SBRT group, P = 0.022 (Table 1).

This study demonstrated that post-RT IDO activity/changes are significantly associated with OS/PFS in stage I-II NSCLC and that RT dose fractionation has a significant effect on IDO-mediated immune activity during- and post-RT in patients with early-stage NSCLC. Our data are congruent with prior studies identifying IDO as a prognostic biomarker in stage III NSCLC (14) and support the hypothesis that hypofractionated RT is less immunosuppressive than more prolonged fractionation schemes.

First, this study validated our previous findings that IDO is a potentially valuable biomarker in stage III NSCLC (14). In the present study, which included only patients with early-stage disease, we showed that fractionated RT induces significant changes in IDO-mediated immune activity, and that IDO immune status after RT correlates with survival outcomes. Similar to stage III NSCLC patients, low IDO activity, as reflected by low kynurenine levels or a low K:T ratio, was significantly associated with improved OS or PFS, as well as low rates of local/regional/distant failure in stage I-II NSCLC patients. This is in agreement with previous studies showing that increased levels of kynurenine or IDO correlated with worse overall survival in various cancers (19–22), including lung cancer (23–25).

Interestingly, as was previously described in stage III NSCLC, baseline kynurenine levels and K:T ratio were not prognostic for either OS or PFS in early-stage NSCLC. To explore the potential mechanism, we conducted an additional exploratory analysis of the relationship between clinical stage and IDO activity in the combined dataset. The results showed that with increasing clinical stage, the kynurenine concentration or K:T ratio decreased; kynurenine levels (2.32 ± 1.25 μmol/L, median ± SD) in stage I patients were significantly higher than in stage III patients (1.86 ± 0.87; P = 0.002; Supplementary Fig. S1). This indicates that IDO-mediated immune status may vary depending on disease stage in patients with NSCLC. For patients with early-stage NSCLC undergoing RT instead of surgery, baseline IDO immune status may not be a decisive prognostic marker. IDO immune status may also reflect frailty in medically inoperable patients with early-stage NSCLC, who generally have serious medical comorbidities and worse overall performance status, which is presumed to be associated with suppressed immune function. Further study is required to better understand the mechanisms underlying the relationship between IDO status and stage in NSCLC.

More importantly, in addition to validating our previous findings that RT can alter IDO-mediated immune activity, this study demonstrated that RT dose fractionation affects IDO-immune status in different ways. In 3DCRT patients, similar to previous work in stage III NSCLC, RT is associated with both favorable and unfavorable changes in IDO immune activity, which vary with dose level (time point) or patient level. During RT, median kynurenine concentration and K:T ratio decreased slightly. The reduction at during-RT suggests that antitumor immunity is enhanced contingently by suppressed IDO activity. Increased antitumor immunity in the TME may be due to RT immunomodulation in the initial phase of RT. Possible mechanisms underlying this finding include decreased tumor burden after massive radiosensitive tumor cell killing by ionizing irradiation or RT-induced enhancement of antitumor immunity in the TME. Although the changes at the during-RT time point were not significant, the trend of reduction was consistent with the findings of our previous study (14), supporting the hypothesis that RT can stimulate anti-tumor immunity, especially in the initial phase of RT. Following RT, IDO activity significantly increased in most 3DCRT patients, suggesting an evolution from an initial immunostimulatory effect of RT to an immunosuppressive phenotype that develops after a prolonged treatment course. This could be due to either proliferation of radioresistant tumor cells, which begin to dominate the TME and suppress antitumor immunity, or that longer courses of RT themselves may lead to depressed anti-tumor immune responses. Therefore, monitoring IDO immune status during RT may assist in determining an optimal individualized RT dose for each patient, which balances RT-induced immunostimulatory and immunosuppressive effects. In SBRT patients, however, changes in IDO biomarkers were not significant either during- or post-RT. During RT, median kynurenine levels decreased, but the median K:T ratio increased slightly. However, a slight increase of IDO activity during RT may not be an indication of unfavorable change; rather, such changes could reflect an acute immune response to normal tissue radiation damage following high-dose radiation in SBRT patients. After RT, the levels of kynurenine and K:T ratio increased consistently, but to a lesser extent than in 3DCRT patients. This suggests that hypofractionated RT may be less immunosuppressive than longer treatment schemes. Besides the biological effect of hypofractionation, differences in RT techniques may also affect IDO dynamics. It is important to note that tumor volumes in the SBRT patients were smaller than the 3DCRT patients (Table 1), and that SBRT is more conformal than 3DCRT, both of which could affect IDO dynamics.

RT dose fractionation effects on antitumor immunity have not been adequately studied previously in the clinic. However, the synergistic effect of RT and immune therapy has been reported in several studies. To obtain optimal synergy, mathematical models have been constructed in an attempt to identify the optimal timing for RT/immunotherapy combinations (26). For example, based on a fitted animal model, Chakwizira and colleagues (27) developed a mathematical model to predict the number of RT fractions needed to maximize the synergistic effect of RT/IDO inhibitor combinations. Our data suggest that RT-induced changes in antitumor immunity exhibit variability based on dose fractionation as well as inter-individual differences. Monitoring IDO-mediated immune status could be a useful and practical way to identify patients with differing antitumor immune status and deliver individualized RT treatment using optimal dose fractionations to obtain the best chance of cancer control. Obviously, a larger prospective study is needed to validate our findings.

Limitations of the present study include serum sample availability. The numbers of available samples at different time points varied, which may affect study results, especially for inter-group comparisons. Also, this small series was not sufficiently powered to permit a subgroup survival analysis for 3DCRT and SBRT. Although most patients did not receive chemotherapy during RT, effects of chemotherapy on the immune system cannot be excluded completely because 5 of the 3DCRT patients in the present series received concurrent chemoradiation.

In conclusion, in addition to validating our previous findings that RT-induced alterations in immune status are reflected in IDO activity, with an initial phase of immunostimulation followed by post-RT immunosuppression, this study demonstrated that RT dose fractionation has a significant effect on IDO-mediated immune status. IDO is a potentially valuable biomarker for monitoring immune status during RT for NSCLC and tailoring treatment and fractionation schedule accordingly.

A. L. Mellor is an employee/paid consultant for Newlink Genetics and holds ownership interest (including patents) in Augusta University. No potential conflicts of interest were disclosed by the other authors.

Conception and design: W. Wang, L. Huang, J.-Y. Jin, F.M. Kong

Development of methodology: W. Wang, L. Huang, J.-Y. Jin, W. Pi, F.M. Kong

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W. Wang, L. Huang, W. Pi, S. Jolly

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W. Wang, L. Huang, J.-Y. Jin, W. Pi, S.G. Ellsworth, S. Jolly, M. Machtay

Writing, review, and/or revision of the manuscript: W. Wang, L. Huang, J.-Y. Jin, W. Pi, S.G. Ellsworth, S. Jolly, A.L. Mellor, M. Machtay, F.M. Kong

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Huang, W. Pi, A.L. Mellor, M. Machtay, F.M. Kong

Study supervision: L. Huang, F.M. Kong

This work was supported by the National Cancer Institute, National Institutes of Health, R01 CA142840 [PI: F.M. Kong].

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.