Tumor molecular profiling

Molecular profiling prior to the initiation of treatment was not required for trial enrollment. Post-hoc analysis of available tumor molecular profiling was performed for the current report. The specific profiling method depended on the availability and quality of appropriate tumor and/or blood samples. Tumor samples, when available, underwent in-house next-generation sequencing (NGS assay, DNA-based panel), capable of detecting 134 gene single nucleotide variants (SNVs) and 47 gene copy number gains, totaling 146 genes performed by the institutional Clinical Laboratory Improvement Amendments-certified Molecular Diagnostic Laboratory at MD Anderson Cancer Center, as previously described.19 20 Additionally, targeted intergenic and intragenic fusions in 51 genes were also identified using an NGS assay (RNA-based panel). By clinician discretion and tissue availability, cytogenetic analyses by fluorescent in-situ hybridization (FISH) were performed to identify MET amplification and rearrangements in RET, ALK, and ROS1. In certain cases, in-house NGS-based circulating cell-free DNA analysis via institutional liquid biopsy was performed using a blood sample for genomic profiling, capable of detecting SNVs in 70 genes, 19 gene copy number gains, and six gene fusions, as described previously.19 21 Tumor cell programmed death-ligand 1 (PD-L1) expression was evaluated on available baseline tumor samples as part of the research correlative analyses, when possible, using single chromogenic immunohistochemistry analysis (clone 28–8, ab205921, 1:100 dilution, Abcam).3 The details of the tissue preparation process have been previously described.22 PD-L1-stained slides underwent scoring by two independent and trained pathologists following the guidelines recommended by the International Association for the Study of Lung Cancer.23 Standard microscopy was used to identify any positive membrane staining on malignant tumor cells and determine the tumor proportion score.23 Tumor tissue molecular profiling reported here, involving NGS-based and/or cytogenetic (FISH-based) testing, was conducted on samples obtained from the primary tumor and/or surgical resection and/or metastatic lesion through bronchoscopic biopsy and/or radiographically guided percutaneous biopsy. Institutional liquid biopsy was obtained, where possible, through blood samples at the discretion of treating physicians at the time of suspected or evident radiographic disease recurrence/progression during surveillance or on tissue confirmation of the presence of disease.

Current study objectives

The purpose of this exploratory report was to describe the impact of select AGAs on TF, MPR and tumor pathological regression in patients treated with neoadjuvant ICIs. Additionally, the study explored the impact of ICI-based therapy, targeted therapy, or chemotherapy, as a subsequent treatment option after failure from neoadjuvant ICIs. TF was defined as the progression of primary lung cancer after neoadjuvant therapy in patients without surgery, radiographic and/or biopsy-proven primary lung cancer recurrence after surgery, or death from possibly treatment-related complications or from primary lung cancer, from randomization on study. The database cut-off date was August 2, 2023.

Characteristics of tumor AGAs and non-AGAs+NP SCC patients

Among the 44 patients, 21 (47.7%) had undergone tumor and/or blood molecular profiling. Of the 23 patients without profiling, 12 had adenocarcinomas and 11 had squamous histology at baseline (figure 2). In an exploratory fashion, we compared patients with AGAs (n=12) to the non-AGAs+NPSCC group (n=20), which consisted of two patient cohorts: Patients who underwent tumor molecular profiling but were found to lack AGAs (non-AGAs, n=9) and patients with NP SCC (n=11) (figure 2).

Open in a separate window

Figure 2

Patients with tumors harboring actionable genomic alterations (AGAs, n=12) were compared to the non-AGAs + NP SCC group, which included profiled tumors without AGAs and non-profiled squamous cell carcinomas (n=20); Tumor histology is at baseline. *Molecular profiling methods included next-generation sequencing (NGS DNA panel and/or NGS RNA fusion panel, when available), and/or cytogenetic studies (when available), and/or circulating cell-free DNA (ccfDNA) (when available) via liquid biopsy. ALK, anaplastic lymphoma kinase; EGFR, epidermal growth factor receptor; n, number; NP SCC, non-profiled squamous cell carcinomas.

AGAs were found in 12 (57.1%) profiled patients. These consisted of five mutations in EGFR (2 Exon 21L858R mutations, 1 Exon 19 deletion with MET amplification, 1 Exon 19 deletion with RET deletion, 1 Exon 20 duplication with MET amplification), 2 KRAS G12C mutations, 1 ERBB2 Y772_A775dup mutation, 1 BRAF V600E mutation, 2 ALK (1 EML4-ALK fusion, 1 rearrangement) fusions and 1 KIF5B-RET fusion. TP53 was the most frequently altered gene, present in 57.1% (12/21) of sequenced samples. The most frequent AGAs were EGFR mutations, which were found in 23.8% (5/21) of sequenced samples, followed by KRAS G12C mutations found in 9.5% (2/21). Of note, there were four STK11 mutations, two of whom occurred with KRAS (1G12C, 1 Q61H) co-mutation. Evaluation of clinicopathological characteristics in AGAs and non-AGAs+NPSCC patient groups revealed that the AGAs group had more patients younger than 65, and more never-smokers. Non-AGA+NPSCC group had a higher number of patients with squamous histology (table 1). Gender, race, clinical stage, and pretreatment tumor PD-L1 expression status were not different between the two groups (table 1).

Table 1

Clinicopathological characteristics of AGAs and non-AGAs+NPSCC patients

	AGAs(n=12)	Non-AGAs + NPSCC(n=20)
Age
< 65	7 (58%)	5 (25%)
≥ 65	5 (42%)	15 (75%)
Gender
Female	4 (33%)	7 (35%)
Male	8 (67%)	13 (65%)
Smoking status
Never smoker	5 (42%)	2 (10%)
Former/current smoker	7 (58%)	18 (90%)
Race
White	9 (75%)	17 (85%)
Non-white	3 (25%)	3 (15%)
Stage
Stage I/II	9 (75%)	15 (75%)
Stage IIIA	3 (25%)	5 (25%)
Histology*
Squamous	1 (8%)	16 (80%)
Non-squamous	11 (92%)	4 (20%)
PD-L1†
<1 %	6 (67%)	8 (67%)
≥1 %	3 (33%)	4 (33%)

Open in a separate window

Demographic information for patients with tumors harboring actionable genomic alterations (AGAs) or patients with tumors not harboring AGAs or non-profiled squamous cell carcinomas (non-AGAs+NP SCC). The percentages for each characteristic are rounded to the nearest whole number to total 100%.

^*Tumor histology at baseline.

^†Pretreatment tumor PD-L1 expression status (n=21) on available samples by clone 28–8, Abcam3 23; the PD-L1 expression status of three patients in AGAs and eight patients in non-AGAs+NPSCC were unavailable.

PD-L1programmed death-ligand 1

Clinicopathological and molecular characteristics of patients with TF

Among the 37 patients who were resected on-trial, 27.0% (10/37) had TF, and among the 7 patients who did not undergo surgical resection or were resected off-trial, 71.4% (5/7) had TF. Out of 15 TFs, 60% (9/15) had identifiable AGAs.

TF was more common in patients with AGAs than in the non-AGAs+NPSCC group (75% (9/12) vs 20% (4/20)). The most common AGAs in the TF group were EGFR mutations (2 Exon 21 L858R mutations, 1 Exon 19 deletion with MET amplification, 1 Exon 19 deletion with RET deletion, 1 Exon 20 duplication with MET amplification), followed by one EML4-ALK fusion, one KIF5B-RET fusion, one KRAS G12C mutation and one BRAF V600E mutation. Six squamous cell carcinomas were profiled, one had an ALK rearrangement, and five did not harbor an AGA.

Among the 32 patients in AGAs and non-AGAs+NPSCC groups, 28 had pathologic response assessment of residual viable tumor in the resected tumor specimen. The odds of achieving MPR in the non-AGAs+NPSCC group were 4.71 (95% CI: 0.49 to 45.2) times higher than in the AGAs group (figure 3A). The AGAs group had a higher median percentage of residual viable tumor than the non-AGAs+NPSCC group (72.5% vs 33.0%) (figure 3A).

Open in a separate window

Figure 3

(A) Violin plot showing the pathological response in the resected tumor specimens from patients who underwent curative-intent surgery on trial in AGAs group (patients with tumors harboring an actionable genomic alterations, n=9) versus the non-AGAs+NP SCC group (patients with genomic profiling without AGAs or with non-profiled squamous cell carcinomas, n=19). The dds ratio of MPR was calculated in univariable logistic regression comparing the odds of achieving MPR in non-AGAs+NP SCC group with AGAs group. The dashed line indicates the median; the dotted lines indicate the lower quartile and upper quartile values; the top and bottom of the violin plots indicate the minima and maxima. (B) Kaplan-Meier curves of treatment failure-free survival probability for patients in the non-AGAs+NP SCC (n=20) and AGAs (n=12) groups. The 95% CIs are shown in parentheses by each treatment failure-free rate at 36, 48, and 60 months. The median time to TF with 95% CI was estimated using Kaplan-Meier method. Treatment failure was defined as any progression of primary lung cancer in patients without surgery, radiographic and/or biopsy-proven lung cancer recurrence after surgery, or death from possibly treatment-related complications or from primary lung cancer, since randomization. AGAs, actionable genomic alterations; ALK, anaplastic lymphoma kinase; EGFR, epidermal growth factor receptor; MPR; major pathologic response (residual viable tumor ≤ 10% in the resected tumor specimen); NE, not evaluable; NP SCC, non-profiled squamous cell carcinomas; TF, treatment failure.

16.7% (2/12) of patients whose tumors had a TP53 mutation had an MPR, and none (0/4) of the tumors harboring an STK11 mutation achieved MPR. Three of the four patients with STK11 mutations were in the non-AGAs+NPSCC group, and one (33.3%) had TF, this patient had a KRAS Q61H co-mutation. The only AGAs patients with MPR harbored a BRAF V600E mutation and developed recurrence in the brain 40.4 months after randomization and was treated with metastasectomy and postoperative radiation and remained disease-free at the time of dataset cut-off.

Patients with tumors harboring AGAs had a shorter median time to TF than those without AGAs. The median time to TF for AGAs patients was 24.7 months (95%CI: 12.6 to 40.4), and was not reached (95%CI: not evaluable (NE)–NE) in the non-AGAs+NPSCC group (figure 3B). TF-free rates at 36, 48, and 60months were 33.3% (95% CI: 10.3% to 58.8%), 22.2% (95% CI: 4.11% to 49.2%), and 22.2% (95% CI: 4.11% to 49.2%) for the AGAs group, and 78.8% (95% CI: 52.8% to 91.5%), 78.8% (95% CI: 52.8% to 91.5%), and 78.8% (95% CI: 52.8% to 91.5%) for the non-AGAs+NPSCC group.

The risk for TF was also assessed for association with molecular subgroup, age, sex, smoking status, race, clinical stage, tumor histology, MPR status (for resected patients on-trial), and pretreatment tumor PD-L1 expression status (figure 4). The risk for TF was significantly higher for AGAs (HR: 5.51, 95%CI: 1.68 to 18.1), and lower in former/current smokers than never-smokers (HR: 0.24, 95%CI: 0.08 to 0.75). 80% (4/5) of never-smokers with TF also had AGAs (2 EGFR, 1 EML4-ALK fusion, and 1 KIF5B-RET fusion). Patients with squamous histology also had a lower risk of TF compared with those with non-squamous histology (HR: 0.20, 95%CI: 0.06 to 0.74). A multivariable Cox PH regression model was performed. However, since collinearity existed in histology and molecular subgroup, only smoking status and molecular subgroup were considered in the model. The AGAs had a higher risk for TF than the non-AGAs+NPSCC (HR: 4.35, 95%CI: 1.24 to 15.2) while the smoking status was no longer associated with TF (HR: 0.41, 95%CI: 0.12 to 1.37). The median time to first systemic retreatment was 20.2 months (range: 10.7–32.0) after randomization. The treatment course over time for non-AGAs+NPSCC (n=20, patients 1–20) and AGAs (n=12, patients 21–32) patients is displayed in figure 5. The specific genomic alterations for the available patients are detailed in online supplemental table 1).

Open in a separate window

Figure 4

Forest plot showing the treatment failure HR for each variable. The HR from the univariable Cox proportional hazard regression estimates the risk of TF in the interest group relative to the risk in the reference group. Treatment failure was defined as any progression of primary lung cancer in patients without surgery, radiographic and/or biopsy-proven lung cancer recurrence after surgery, or death from possibly treatment-related complications or from primary lung cancer, since randomization. * Data collected at time of surgery in resected tumors in patients who underwent surgery on trial (n=28), three patients in AGAs and one patient in non-AGAs+NP SCC did not undergo surgery; time from surgery until treatment failure was used. ** Pretreatment tumor PD-L1 expression status (n=21) on available samples by clone 28-8, Abcam3 23; the PD-L1 expression status of three patients in AGAs and eight patients in non-AGAs+NP SCC were unavailable. § Tumor histology at baseline. AGAs, patients with tumors harboring an actionable genomic alteration; MPR, major pathologic response (≤ 10% residual viable tumor in the resected tumor specimen); non-AGAs+NP SCC, patients with molecular profiling without AGAs or non-profiled squamous cell carcinomas; PD-L1, programmed death-ligand 1; TF, treatment failure; HR: hazard ratio.

Open in a separate window

Figure 5

Swimmer plot showing duration of time free from treatment failure (blue) and time since treatment failure (red) in patients with tumors without an actionable genomic alteration (AGAs) or a non-profiled squamous cell carcinomas (non-AGAs+NP SCC, patients 1–20) and patients with tumor harboring an AGA (AGAs, patients 21–32). Chemotherapy, ICI-based therapy, and targeted therapy were started after treatment failure (TF). Non-profiled adenocarcinomas (n=12) are not included. Type of treatments each patient received are displayed chronologically based on time since randomization. Patient 23 had surgery off-trial. Patient 24 did not have disease progression after neoadjuvant therapy, was offered curative-intent surgery but declined. ALK, anaplastic lymphoma kinase; EGFR, epidermal growth factor receptor; ICI, immune checkpoint inhibitor.

A sensitivity analysis was performed by assigning the two patients with KRAS G12C to the non-AGAs+NPSCC group. The odds of achieving MPR in the non-AGAs+NPSCC + KRASG12C group were 3.38 (95% CI: 0.35 to 32.6) times higher than in the AGAs group (online supplemental figure 1A). The AGAs group had a higher median percentage of residual viable tumor than the non-AGAs+NPSCC + KRASG12C group (61.0% vs 34.5%) (online supplemental figure 1A), AGAs had a median time to TF of 19.2 months (95%CI: 2.57 to 40.4), and was not reached (95%CI: NE–NE) in non-AGAs+NPSCC + KRASG12C group (online supplemental figure 1B). The TF risk in the AGAs group was 5.96 (95% CI: 1.92 to 18.5) times higher than the non-AGAs+NPSCC + KRASG12C group.

Retreatment with ICI-based therapy

Among the 13 patients with TF and retreatment, 4 were retreated with ICI-based therapy, and 1 patient had disease control. This patient experienced disease recurrence 18.5 months after randomization and was treated with a chemoimmunotherapy regimen with stable disease for 2.1 months. Tumor profiling revealed TP53 and EGFR L858R Exon 21 mutations while on ICI-based therapy and the treatment was subsequently changed to osimertinib with disease control persisting 38.5 months after the first disease recurrence. The remaining three patients treated with ICI-based therapy did not achieve disease control. One patient recurred at 12.6 months and was treated with chemoimmunotherapy without response, while molecular profiling was pending. Molecular testing showed EGFR Exon 20 duplication with MET amplification and this patient was subsequently treated with targeted therapy with disease control for 6.20 months before radiographic disease progression. At the time of disease progression, treatment was changed to immunotherapy, but the patient died from lung cancer-related complications 2.47 months later. Another patient underwent definitive chemoradiation after neoadjuvant therapy (no surgery) and later experienced locoregional disease progression. Molecular profiling failed to reveal any AGAs and the patient started single-agent immunotherapy, but unfortunately died 4.80 months after treatment from cancer-related complications. Finally, one patient with recurrence 17.0 months after randomization was treated with one cycle of chemoimmunotherapy without response. The molecular profiling revealed a KIF5B-RET fusion, and the patient was enrolled in a RET TKI trial. However, the patient died from disease-related complications 4.67 months after retreatment initiation.

Treatment with targeted therapy alone

In addition to the patients treated with ICI-based therapy and targeted therapy described above, four additional patients were treated with targeted therapy alone. One patient with EML4-ALK fusion had metastatic recurrence at 20.0 months after randomization and was treated with targeted therapy with disease control at 18.5 months until disease progression. One patient with EGFR Exon 19 deletion was treated with targeted therapy with disease control at 22.8 months with ongoing treatment. One patient developed recurrence at 15 months after randomization and had EGFR L858R Exon 21 mutation, the patient was treated with targeted therapy with ongoing disease control at 27.5 months. Finally, one patient with EGFR Exon 19 deletion and MET amplification, had disease progression after neoadjuvant treatment, and was treated with definitive chemoradiation (no surgery). The disease eventually metastasized, and the patient was treated with targeted therapy, but ultimately died from complications secondary to lung cancer 6.37 months after targeted therapy initiation. In total, disease control was achieved with targeted therapy in 71.4% (5/7) of patients with TF who received targeted therapy alone or targeted therapy given in sequence with other systemic therapies.

We thank the patients and their families for participating in this study. We thank all the members of our regulatory, clinical, data coordination and translational research teams in the Departments of Thoracic/Head and Neck Medical Oncology, and Thoracic Surgery at the MD Anderson Cancer Center for their support on this study. We thank the members of the Translational Molecular Pathology Department involved in the correlative analyses of this study and members of the strategic alliance teams at Bristol Myers Squibb and the MD Anderson Cancer Center for their support.