Search

Purpose Consolidative thoracic radiation therapy (TRT) has been shown to improve outcomes for patients with extensive stage small cell lung cancer. We hypothesized that the addition of ipilimumab (IPI) and nivolumab (NIVO) after TRT would improve outcomes for patients with extensive stage small cell lung cancer. Methods and Materials Eligibility required stable disease or better after platinum doublet chemotherapy. Study therapy included consolidative TRT to 30 Gy in 10 fractions, targeting residual primary tumor and initially involved regional lymph nodes. Two weeks after TRT, patients received concurrent IPI (3 mg/kg) and NIVO (1 mg/kg) every 3 weeks for 4 doses followed by NIVO monotherapy (480 mg) every 4 weeks until progression or up to 1 year. Results The study enrolled 21 patients, with 6-month progression-free survival (PFS) of 24% (90% confidence interval [CI], 11%-40%) and a median PFS of 4.5 months (95% CI, 2.7%-4.6%). The 12-month overall survival (OS) was 48% (95% CI, 29%-64%) with a median OS of 11.7 months (95% CI, 4.7%-16.0%). Fifty-two percent of patients had ≥1 possibly related grade 3 to 4 immune-related adverse event. Grade 3 pulmonary and gastrointestinal immune-related adverse events were recorded in 19% and 24% of patients, respectively. Exploratory analysis showed increased cytotoxic T cell (CD3+CD8+) tumor infiltration was associated with favorable PFS (P = .01) and OS (P = .02). Reduction in peripheral blood CD3+CD8+ from baseline to after first dose of IPI/NIVO was associated with improved PFS (P = .02) and OS (P = .02). Conclusions Consolidative IPI and NIVO after platinum-based chemotherapy and TRT demonstrated a toxicity profile consistent with the known adverse events attributable to IPI and NIVO. Although the study regimen did not significantly improve PFS, the OS was higher than historic expectations. CD3+CD8+ tumor infiltration and migration may identify patients most likely to have improved outcomes in small cell lung cancer.

Purpose/Objective(s) We hypothesized that the addition of concurrent ipilimumab (IPI) with chemoradiotherapy followed by consolidative nivolumab (NIVO) would be safe and tolerable for patients with unresectable stage III non-small cell lung cancer (NSCLC). We report early outcomes and toxicity associated with this regimen in a phase I/II clinical trial. Materials/Methods The study was designed as a prospective, multicenter phase I/II trial. Eligible patients had ECOG 0-1 and unresectable stage III NSCLC. Therapy included a platinum-based chemotherapy doublet with concurrent radiotherapy to 60 Gy in 30 fractions over six weeks and two doses of concurrent IPI (1 mg/kg) given weeks 1 and 4. Maintenance NIVO was initiated 1-3 weeks after completion of chemoradiotherapy/IPI and given every 4 weeks (480 mg) for up to 12 cycles. The primary endpoints were to evaluate the safety and tolerability of the regimen (Phase I) and the 12-month PFS (Phase II). Treatment-related toxicity was assessed according to CTCAE v5.0. Time to event analysis was performed with Kaplan Meier (KM) and Cox proportional hazard (CPH) models. Results are reported with 95% confidence intervals (CI). Results Nineteen of a planned 55 patients were enrolled in the trial which was discontinued without proceeding to the Phase II component due to exceeding the futility boundary for toxicity. The median follow-up was 21 months by the reverse KM method. The 12-month PFS estimate was 54% (CI 29-78) and the median PFS was 19 months (CI 6-not reached). The 12-month OS estimate was 60% (CI 36-84) while the median OS was not reached. Ten patients (53%, CI 29-76) experienced grade 2 or higher (G2+) pneumonitis. The median time to development of G2+ pneumonitis was 5.5 months and the risk of G2+ pneumonitis at 6- and 12-months was 57% (CI 30-84) and 74% (CI 49-99), respectively. Sixteen patients (84%, CI 68-100) had any G3+ treatment related toxicity. The most common G3+ toxicities were pulmonary (8 patients, 42%, CI 20-67) and cytopenias (7 patients, 37%, CI 16-62). Five patients (26%, CI 6-46) had possible treatment related G5 toxicity, including three patients with possible treatment related G5 pulmonary toxicity (16%, CI 0-32). There was no difference in the mean percent of lung volume receiving 20 Gy (V20) between those with and without G2+ pneumonitis (26%, CI 20-32 vs 21%, CI 16-27, P = 0.18), nor in the mean lung dose (14 Gy, CI 10-17 vs 15 Gy, CI 12-18, P = 0.35). Neither mean lung dose nor lung V20 were associated with time to G2+ pneumonitis on univariable CPH. Conclusion The combination of concurrent IPI with standard chemoradiation followed by maintenance NIVO for unresectable stage III NSCLC is associated with significant toxicity which may limit opportunities for improved outcomes, although the sample size in this trial is small. Alternative strategies or sequencing should be explored to integrate immunotherapy with cytotoxic chemotherapy and radiation for patients with unresectable stage III NSCLC.

Purpose/Objective(s) Immune checkpoint inhibitors (ICI) and epidermal growth factor receptor-tyrosine kinase inhibitors (EGFR-TKI) have shown responses in the management of non-small cell lung cancer (NSCLC) brain metastases (BM). However, optimizing control with stereotactic radiosurgery (SRS) and various systemic therapies in the management of NSCLC BM remains an area of investigation. We evaluated clinical outcomes in NSCLC BM treated with SRS and ICI, EGFR-TKI, ICI and chemotherapy, compared to conventional chemotherapy. Materials/Methods A total of 174 NSCLC BM patients, treated between 2015 and 2019, to a total of 660 intact (non-operative) BM over 239 single-fraction SRS sessions within 3 months of receiving ICI (n = 58; 33%), EGFR-TKI (n = 30; 17%), chemotherapy and ICI (n = 22; 13%) or standard chemotherapy (n = 64; 37%) were identified. Time-to-event analysis was conducted with the Kaplan-Meier method, and outcomes included distant intracranial control (DIC), local control (LC), and overall survival (OS) from SRS. Cox proportional hazards model was utilized for multivariate analysis (MVA) to identify correlations of clinical variables with outcomes. Results Median follow-up from BM diagnosis was 8.9 months (0.26-126.5 months) and 7.3 months from SRS (2.5-17.6 months). No significant differences in age (P = 0.55), gender (P = 0.49), treated gross tumor volume (P = 0.31), presence of systemic metastasis (P = 0.1), SRS dose (P = 0.61), or KPS (0.35) were noted between treatment groups. The 12-month DIC was 35%, 53%, 41%, and 20% (P = 0.02) for ICI, EGFR-TKI, ICI and chemotherapy, and chemotherapy alone groups, respectively. No differences were noted in LC (P = 0.1) and OS (P = 0.5) between treatment groups. Upon MVA, factors found to be significant for improved DIC included treatment with EGFR-TKI therapy compared to conventional chemotherapy (HR 0.4; 95% CI 0.25-0.76; P = 0.04) and treatment with SRS before systemic therapy (HR 0.6; 95% CI 0.3-0.9; P = 0.03). Factors found to be significant for improved LC on MVA included treatment with SRS before (HR 0.4; 95% CI 0.2-0.9; P = 0.03) or concurrent (HR 0.3; 95% CI 0.1-0.6; P = 0.003) compared to following receipt of systemic therapy. Rates of radiation necrosis (RN) did not differ between treatment groups and were 6% (n = 12), 5% (n = 12), 3% (n = 3), 3% (n = 3), for ICI, EGFR-TKI, ICI and chemotherapy, and chemotherapy alone groups, respectively, P = 0.43. In addition, no differences were noted in RN between timing of SRS and receipt of systemic therapy, (P = 0.81). Conclusion In our analysis of NSCLC BM patients treated with single session SRS alongside systemic therapy, we found significant differences in DIC based on receipt of systemic therapy. In addition, treatment with SRS before or concurrent with systemic therapy improved control rates. No differences in RN were noted between treatment groups or by timing of SRS and systemic therapy. Prospective evaluation of the potential synergistic effect between systemic therapy and SRS in NSCLC BM management is warranted. Author Disclosure A.E. Dohm: None. J. Tang: None. M.N. Mills: None. B.A. Perez: Research Grant; BMS.T.J. Robinson: None. B. Creelan: Research Grant; NGenomics Laboratories, Adaptive Biotechnologies Corp. Consultant; Xilio Inc, E.R. Squibb LLC, F. Hoffmann-La Roche AG, AstraZeneca plc, AbbVie, KSQ Therapeutics Inc, GlaxoSmithKline plc, Gilead Sciences Inc, Celgene corp. Speaker's Bureau; AstraZeneca plc, ARIAD Pharmaceuticals, Hoffmann-La Roche AG.J. Gray: None. A.B. Etame: None. M. Vogelbaum: Honoraria; Celgene, Tocagen. Stock Options; Infuseon Therapeutics. Royalty; Infuseon Therapeutics. P. Forsyth: Research Grant; Pfizer, Celgene. Advisory Board; Novocure, BTG, Inovio, AbbVie, Ziopharm, Tocagen, Pfizer. H.M. Yu: Honoraria; UpToDate. Speaker's Bureau; BrainLab. Advisory Board; Novocure, AbbVie. D.E. Oliver: None. K.A. Ahmed: Research Grant; Bristol-Myers Squibb, Genentech.

Purpose/Objective(s) Prescription dose for lung SBRT remains a controversial topic. Based on previous experience, a BED of 100 Gy is necessary to maximize tumor control, though many have advocated higher prescription doses (or dosimetric “hot spots”). To analyze this "imbalanced" problem (few events) and identify which factors affect not only events but the time leading to these events, we hypothesized that advanced actuarial (survival) analysis with machine learning techniques across the entirety of the IGTV/PTV DVH curves in a very large data set would lead to novel findings. Materials/Methods We evaluated 903 courses of lung SBRT, all prescribed to 50 Gy in 5 fractions (BED = 100). There were 37 local failures (crude rate = 4.1%) with median time-to-failure of 587 days (range: 96-1777). Variables analyzed included patient characteristics (gender, age), tumor features (maximum diameter, IGTV/PTV volumes, tumor site, stage), and IGTV/PTV DVH statistics: V1, V2, ... Dmax; D5, D10, ... D95; D0.03cc; mean/median/standard deviation dose; and area under the curve. Mean number of variables per SBRT course was 184 (nearly all dosimetric). Actuarial machine learning methods using Elastic Net with Cox regression (ENC) and Random Survival Forest (RSF) with right censoring were applied in the analysis. ENC is a weighted combination of ridge (penalized Cox with L2 norm) and LASSO (penalized Cox with L1 norm). RSF is an ensemble of survival decision trees constructed by bagging of bootstrapped samples and provides higher predictive power compared to traditional ENC. Categorical variables were represented using one-hot encoding. The data was split 75:25 for training and testing, respectively. Cross-validation was used to mitigate overfitting pitfalls. Variables’ importance was assessed using feature permutation analysis. Results ENC identified IGTV volume as the primary predictor of local failure. Conversely, RSF found PTV volume to be most predictive (with IGTV volume also playing a role). RSF also identified IGTV minimum dose, PTVD15, D0.03cc of the IGTV and PTV, PTV standard deviation, and IGTV D50 and V50 as important features. This combination of features suggests that a minimum BED of 100 Gy is relevant to local control (IGTV minimum dose, IGTV V50), but also that the “hot spot” plays a role (PTVD15, D0.03cc, IGTV D50). We are currently conducting further analysis of these variables to derive dosimetric parameters to incorporate into treatment planning. Conclusion Actuarial machine learning methods such as RSF can identity relevant important variables from highly imbalanced data with few events as in the case of SBRT local failure, while traditional statistical methods like ENC fail to capture such high order features. Our results indicate that in addition to tumor volume, dosimetric variables related to both minimum IGTV dose as well as to maximum IGTV/PTV dose play important roles in SBRT tumor response.