Treatment Intensification in Locally Advanced/Unresectable NSCLC Through Combined Modality Treatment and Precision Dose Escalation

Abstract

The best survival for patients with unresectable, locally advanced NSCLC is currently achieved through concurrent chemoradiation followed by durvalumab for a year. Despite the best standard of care treatment, the majority of patients still develop disease recurrence, which could be distant and/or local. Trials continue to try and improve outcomes for patients with unresectable NSCLC, typically through treatment intensification, with the addition of more systemic agents, or more radiation dose to the tumor. Although RTOG 0617 showed that uniform dose escalation across an unselected population of patients undergoing chemoradiation is not beneficial, efforts continue to select patients and tumor subsets that are likely to benefit from dose escalation. This review describes some of the ongoing therapeutic trials in unresectable NSCLC, with an emphasis on quantitative imaging and precision radiation dose escalation.

Introduction

The standard of care for unresectable locally-advanced non-small cell lung cancer (LA-NSCLC) was concurrent chemoradiation for more than a decade¹, until the PACIFIC trial showed that adding a year of adjuvant durvalumab improved survival for patients post-chemoradiation^{2, 3}. Prior to the PACIFIC trial results, many trials tested intensifying systemic therapy through the addition of either induction or consolidation chemotherapy to the concurrent chemoradiation treatment backbone, without encouraging results. Attempt to intensify radiation treatment through dose escalation in RTOG 0617 also did not yield an improvement in survival^{4, 5}. In this review, we examine ongoing efforts to improve survival in LA-NSCLC through additional systemic therapy, usually in the form of more checkpoint inhibitor therapy, and efforts to escalate radiation dose utilizing strategies that are distinct from the uniform dose escalation used in RTOG 0617. We highlight the role of quantitative imaging in dose-escalation strategies to achieve precision radiation therapy.

Intensifying Systemic Therapy

Immunotherapy

After concurrent chemoradiation was established as standard of care for unresectable, LA-NSCLC, a number of trials attempted to improve patient survival by adding chemotherapy before or after concurrent chemoradiation, none with a clear benefit to survival^{6, 7}. However, the PACIFIC trial established 1-year of consolidation durvalumab after chemoradiation as the new standard of care for patients treated with definitive chemoradiation, by showing improved OS and progression-free survival (PFS), with tolerable toxicity^{2, 3}.

Other PD-1/PD-L1 pathway checkpoint inhibitors are also being tested as consolidation therapy after chemoradiation, as outlined in Table 1, including pembrolizumab, and nivolumab. Beyond adding PD-1/PD-L1 inhibitors as consolidation after chemoradiation, trials are also underway to explore adding them during chemoradiation, and as neoadjuvant therapy prior to chemoradiation, as summarized in Table 1.

Table 1.

Clinical trials testing addition of PD-1/PD-L1 inhibitors to concurrent chemoradiation in LA-NSCLC.

Trial Name	Drug	Phase	Patient No.	Results/Comments
PD-1/PD-L1 inhibitors as consolidation after chemoradiation
PACIFIC2, 3	Durvalumab	3	713	18-month PFS=44.2% with durvalumab, 27% with placebo
LUN 14–17950	Pembrolizumab	2	93	18-month PFS=49.5%
BIG10CRC LUN 16–08151	Nivolumab +/− ipilimumab	2	20, ongoing	No unexpected safety signals in the first 20 patients
PACIFIC 6	Durvalumab	2	Ongoing	For patients receiving sequential chemotherapy and radiation, testing 2 years of consolidation durvalumab
PD-1/PD-L1 inhibitors concurrent with chemoradiation
PACIFIC 2	Durvalumab	3	Ongoing	Comparing concurrent chemoradiation alone (without consolidation durvalumab) versus chemoradiation plus durvalumab during and after chemoradiation
ECOG EA5181	Durvalumab	3	Ongoing	All patients receive concurrent chemoradiation plus consolidation durvalumab (i.e. PACIFIC regimen), randomizing between concurrent durvalumab or not during chemoradiation
DETERRED52	Atezolizumab	2	40, ongoing	No increase in toxicity for first 40 patients.
NICOLAS53	Nivolumab	2	82, ongoing	No unexpected adverse events or increased toxicities were observed.
CheckMate73L	Nivolumab +/− ipilimumab	3	Ongoing	Compares the current standard of care treatment (chemoradiation then durvalumab) against nivolumab given concurrent with chemoradiation, then either nivolumab alone or with ipilimumab as consolidation therapy.
KEYNOTE-799	Pembrolizumab	2	Ongoing	1 cycle of pembrolizumab and chemotherapy prior to starting radiation with cycle 2 of systemic therapy (chemotherapy and pembrolizumab every 3 weeks). Patients also receive 14 additional cycles of consolidation pembrolizumab every 3 weeks.
PD-1/PD-L1 inhibitors as neoadjuvant therapy prior to chemoradiation
AFT-16	Atezolizumab	2	Ongoing	4 cycles of induction atezolizumab prior to chemoradiation, and for a year as consolidation.
KEYNOTE-799 (also listed above)	Pembrolizumab	2	Ongoing	See description above. 1 cycle of pembrolizumab and chemotherapy prior to starting radiation with cycle 2 of systemic therapy.

PFS=progression free survival.

Beyond checkpoint inhibitors, other forms of immunotherapy have also been explored in the unresectable NSCLC population. The phase III START trial tested tecemotide (a MUC1 antigen specific vaccine immunotherapy, targeting the MUC1 glycoprotein which is overexpressed and abnormally glycosylated in NSCLC and other cancers) as consolidation after chemoradiation (concurrent or sequential) and saw no significant difference in OS compared with placebo. Unplanned subgroup analysis showed a significant improvement in OS for patients who received previous concurrent chemoradiotherapy median OS 30·8 months (95% CI 25·6–36·8) compared with 20·6 months (17·4– 23·9), adjusted HR 0·78, 0·64–0·95; p=0·016). These results have not been confirmed with subsequent clinical trials.

Biologically Targeted Agents

EGFR Tyrosine Kinase Inhibitors (TKI)

In addition to immunotherapy, there has also been a dramatic increase in targeted therapy options for driver mutations in NSCLC, such as EGFR mutations, however the role of these agents in the setting of LA-NSCLC remains to be determined. Multiple trials have been conducted in the unresectable stage III setting, in unselected patient populations, with negative results. SWOG S0023 looked at consolidation docetaxel and erlotinib (EFGR-TKI) after chemoradiation, and showed inferior results with the consolidation therapy, in a population that was unselected for EGFR mutation⁸. RTOG 0617 demonstrated that adding cetuximab in unselected patients also failed to provide a benefit⁴. Adding EGFR inhibitors to concurrent chemoradiation increases toxicity, such as an increase in grade 3 or worse adverse effects from 70% to 86% with cetuximab in ROTG 0617⁴, but there are conflicting data from small trials on whether this would be safe with agents like gefitinib and erlotinib^{9, 10}. This concept continues to be explored in trials that now select for molecular status, such as the phase III LAURA trial (NCT03521154), looking at addition of osimertinib following chemoradiation compared to placebo for patients with EGFR mutation positive stage III unresectable NSCLC. There is also a phase II trial in Japan looking at gefitinib with concurrent chemoradiation in unresectable patients with EGFR mutations¹¹.

Other Biological Agents

Besides EGFR TKIs, TKIs that target the vascular endothelial growth factor (VEGF) pathway have also been explored with concurrent chemoradiation. However, reports of excess toxicity with tracheoesophageal fistula formation have prevented large scale clinical trials with this combination¹².

Targeting the P-I-3 Kinase pathway has been of interest due to the importance of these pathways for DNA repair. Nelfinavir is an HIV protease inhibitor that inhibits P-I-3 kinase dependent pathways in lung cancer cell lines. A phase I/II trial of nelfinavir with concurrent chemoradiotherapy demonstrated good tolerability and a promising median survival of 41 months¹³. This approach will require testing in the phase III setting with the current standard of adjuvant durvalumab in order to determine the value of nelfinavir in the current era of immunotherapy.

Poly (ADP-ribose) polymerase (PARP) inhibitors suppress a cell’s ability to repair DNA damage and could work synergistically with radiation therapy, which causes DNA damage. Veliparib, an oral PARP1/2 inhibitor, has been tested in multiple phase I trials and reported to have tolerable toxicity with concurrent chemoradiation in stage III NSCLC (M14–360/AFT-07, SWOG S1206)^14,15. Phase 2 testing has stagnated after results of PACIFIC were published in 2018 showing the efficacy of checkpoint inhibition.

Radiation Dose Escalation

Current Dilemma with Radiation Dose Escalation

Prior to the results from RTOG 0617 showing inferior survival with concurrent chemoradiation to 74 Gy versus 60 Gy^{4, 5}, multiple retrospective analyses showed an improvement in patient outcomes with radiation dose escalation in unresectable NSCLC. A retrospective analysis of 1356 patients on seven RTOG trials saw a strong association between the biologically effective dose (BED) and both OS and local-regional failure¹⁶. A 1-Gy BED increase in dose was statistically significantly associated with approximately a 4% relative improvement in survival. Phase I/II trials of concurrent chemoradiation to 74 Gy showed promising results with encouraging survival and tolerable toxicity ^{17, 18}. Therefore, the results of RTOG 0617 showing inferior outcomes with 74 Gy was against expectations and reset the standard of care radiation dose to around 60 Gy⁴. However, local control remains a problem in unresectable NSCLC with local failures of around 30% by 2-years and rising to 50% by 3–5 years^{16, 19}. There are multiple theories regarding the lack of benefit with 74 Gy versus 60 Gy in RTOG 0617, including the longer treatment time on the 74 Gy arm (37 fractions versus 30 fractions), and higher dose to normal tissues being detrimental (esophagitis grade and heart dose were both highly correlated with survival)⁴. Since uniform dose escalation in an unselected group of patients seems detrimental to survival, multiple clinical trials are underway to test other strategies for dose escalation, to improve cancer control and survival.

Precision Dose Escalation

FDG PET-Guided Dose Escalation

Multiple dose escalation strategies utilize FDG PET imaging to guide treatment volume. Published data suggests that FDG avidity may be predictive for local recurrences^{20, 21}, both on the pre-chemoradiation scan and the mid-chemoradiation scan. Lung tumors also exhibit varying degrees of tumor response during treatment, ranging from progression to rapid tumor shrinkage. This had led to a mix of approaches , with some investigators utilizing the pre-treatment PET to dose escalate sub-volumes, and some utilizing the mid-treatment PET to adapt treatment volumes. A number of select studies are summarized below.

Based on pre-treatment PET, the phase II PET-boost trial gave ≥72 Gy in 24 fractions to the primary tumor and 66 Gy to involved nodes²². There was a randomization to either boosting the entire primary planning target volume (PTV) or only boosting regions with SUV_max ≥50%. Maximum boost dose for each patient was limited by normal tissue constraints (concept of “isotoxicity”). Chemotherapy was optional. Reported toxicity rates seem higher than conventional chemoradiation although within predefined stopping rules (grade 3+ dysphagia 14.3%, 9 of 107 patients developed fatal pulmonary hemorrhage or esophageal fistula). Cancer control rates have not been reported. It is unclear if the toxicity rates reported may be related to the hypofractionated regimen rather than PET-based dose escalation.

Another trial design based on pre-treatment PET is an ongoing Canadian Phase II randomized trial NCT02788461, comparing conventional chemoradiation (60 Gy in 30 fractions) against metabolically adaptive chemoradiation, where FDG PET-avid tumor sub-volumes receive an integrated boost dose to a maximum of 85 Gy in 30 fractions²³. Boost is given to regions with SUV > 50% SUV_max, with max dose 85 Gy, while respecting normal tissue constraints, and ensuring the conventional PTV receives at least 60 Gy. The Danish NARLAL 2 trial (NCT02354274) is a phase 3 randomized trial comparing standard of care 66 Gy in 33 fractions with a uniform dose distribution, to an inhomogeneous dose distribution determined by the most active (FDG-PET criteria) area of the tumor, with mean doses up to 95 Gy in 33 fractions to the high FDG uptake volumes.

Other trials are testing adaptive/mid-treatment volume-based dose escalation. Kong et al. published on a phase 2 trial of dose escalation up to 86 Gy in 30 fractions, based on a mid-treatment (40–50 Gy) FDG PET²⁴. Pre-treatment PTV received at least 50 Gy and pre-treatment clinical target volume (CTV) received at least 60 Gy, while the mid-treatment PET defined tumor volume received at least 70 Gy. Radiation plans were individualized to a fixed risk of normal tissue toxicity. The trial saw an encouraging 2-year infield local control of 82% and local regional control of 62%. Median survival was 25 months. RTOG 1106 is a randomized phase 2 trial testing this concept, but results are not yet available. Beyond adapting the boost volume based on mid-treatment changes, there are also trials examining selecting only a portion of patients for dose escalation. Since some patients have durable cancer control after 60 Gy of conventional radiation, not all patients will benefit from dose escalation. For example, ongoing phase II trial FLARE-RT (NCT02773238) is selecting patients for dose escalation based on mid-treatment response as assessed by PET, and only non-responders undergo radiation dose escalation (74 Gy) for the second half of chemoradiation²⁵.

Other Strategies for Biology-Guided Dose Escalation

Besides FDG PET, other imaging tracers have been explored for radiation dose escalation, with hypoxia being the most commonly studied. Hypoxia confers radiation resistance, since the primary mechanism of action for radiation therapy is the creation of reactive oxygen species that subsequently damages DNA by generating double strand breaks. Hence, there is a strong biological rationale to boost hypoxic tumors/tumor regions. The RTEP5 study reported on 21 patients with NSCLC, with ¹⁸FDG- and ¹⁸F-miso PET/CT pre-radiation, and ¹⁸FDG PET/CT mid-radiation (42Gy)²⁶. Different boost plans could be designed based on each of the image sets, and the optimal scan and boost strategy is unclear pending more clinical data. A number of trials evaluating FDG PET/CT based dose escalation also have optional hypoxia imaging, including RTOG 1106 with FMISO imaging in a subset of patients, and the PET-boost trial with ¹⁸F-HX4 imaging.

Beyond imaging, there have been studies looking at other biological predictors of radiation dose response. Scott et al. published on a genome-based model for adjusting radiotherapy dose (GARD) across a range of cancer types, including lung cancer²⁷. Expression of ten radiosensitivity index genes (AR, c-Jun, STAT1, PKC-beta, RelA, cABL, SUMO1, PAK2, HDAC1, and IRF1) were combined to calculate a radiosensitivity index (RSI) for each patient. Individual GARD were derived from the radiation dose/fractionation for each patient, the linear-quadratic model, and the individual RSI. A high GARD value was associated with longer 5-year metastasis-free survival in a cohort of breast cancer patients. On multivariate analysis, GARD was significantly associated with local control in the lung cancer cohort. Such a model may allow individualization of radiation treatment doses, based on inherent tumor radiosensitivity as defined by the gene expression panel.

Quantitative Imaging in Precision Radiation Dose Escalation

The promise of precision oncology requires tools to risk stratify patients, target high risk disease with tailored treatments, and assess early treatment response in order to rapidly adapt therapies. Quantitative imaging, the extraction of quantifiable features from medical images, can guide anatomic and biological targeting with radiation therapy to enable more precise dose escalation.²⁸

Target definition for radiation dose escalation

Definition of high-risk targets for dose escalation in LA-NSCLC based on quantitative imaging imposes certain requirements and tradeoffs among the following: image contrast-to-noise ratio, spatial resolution, sensitivity, specificity, reproducibility, and prognostic / predictive power. These properties are affected by the quantitative imaging modality and contrast agent / biological tracer of interest, and may differ based on disease biology. For example, FDG PET/CT is routinely used for imaging LA-NSCLC due to its high image contrast-to-noise ratio in primary tumors, high sensitivity to detect nodal disease, low inter-observer variability,²⁹ reproducibility early during treatment,³⁰ spatial correlation with subsequent local disease recurrence patterns,^{31, 32} and prognostic power of quantitative imaging biomarkers for discriminating survival outcomes based on early response assessment.^{30, 33–35} However, FDG PET/CT has limited spatial resolution relative to contrast-enhanced CT or MRI, and lacks specificity for distinguishing inflammation from active disease. Hypoxia imaging with FMISO PET/CT provides greater specificity related to tumor biology and radioresistance than FDG PET/CT, yet suffers from lower image contrast and more variability in defining targets,³⁶ often requiring manual contouring aided by high resolution CT.

To improve the reproducibility of high-risk treatment targets, patients should be immobilized and scanned in treatment position, ideally with the same motion management strategy utilized during radiation simulation, such as respiratory correlated imaging. For these reasons, targets defined on 4D PET/CT simulation images for dose escalation will be more reproducible than those defined on diagnostic PET/CT images.³⁷ PET images should be corrected for attenuation with anatomically consistent CT images,³⁸ along with enhanced data corrections that improve image contrast, spatial resolution, and quantitative accuracy, all of which are routinely available on the latest generation PET/CT scanners. While PET/MR imaging presents technical advantages over PET/CT imaging for certain disease sites, including central nervous system / abdominal / pelvic / musculoskeletal cancers, its use in target definition of LA-NSCLC is discouraged due to motion artifacts and unreliable attenuation correction. There is ongoing research into the aggregation of imaging biomarkers from one or more imaging modalities to generate radiomic signatures of increased treatment failure risk. These radiomic signatures could potentially map high-risk regions to form radiomic target volumes and associated radiation boost subvolumes.

Image-guided adaptive radiation therapy

Image-guided adaptation of radiation therapy for LA-NSCLC could allow treatment plan adjustments for morphologic changes based on anatomic imaging (i.e. CT), or adjustments for biological changes on functional imaging (i.e. FDG PET). Morphologic changes include gross tumor tissue regression, lung collapse, and pleural effusions, which can trigger re-optimization of radiation dose to existing clinical / planning target volumes or to define adaptive targets. Shrinking radiation fields based on morphologic changes poses risk of undertreating subclinical disease in regions previously comprised of gross disease. The frequency of anatomic adaptation varies based on tumor biology, radiation target location, target volume size / shape, and radiation modality. Photon VMAT plans with multiple arcs have dose distributions that are typically not significantly impacted by tissue density changes, whereas proton plans with few fields are more sensitive to tissue density changes. The frequency of imaging guidance (daily, weekly, etc.) can also impact when plan adaptation is triggered.

While anatomically adaptive radiation therapy seeks to tailor radiation dose to dynamically evolving target sizes and shapes, biologically adaptive radiation therapy seeks to incorporate tumor biology and spatially variable radiation sensitivity into the treatment plan. There are various approaches to treatment plan optimization³⁹ that could incorporate variables such as FDG avidity, calculated tumor control probability, normal tissue complication probability, or fixed integral dose redistribution.⁴⁰ Figure 1 shows examples of anatomic and biologic adaptive radiation treatment plans, and an example of heterogeneous dose escalation based on FDG avidity.

Figure 1.

Precision image-guided adaptive radiation therapy and selective dose escalation for locally advanced non-small cell lung cancer. (A) FLARE-RT protocol (NCT02773238) patient with responsive primary tumor on Week 3 mid-chemoradiation FDG PET/CT fusion (hot metal colorwash) relative to baseline, requiring anatomically adaptive treatment plan to maintain target coverage at standard dose of 60 Gy in 30 fractions. (B) FLARE-RT protocol patient with non-responsive primary tumor on Week 3 FDG PET/CT. (C) Biologically adaptive dose escalation to increase local control using a dose painting-by-numbers paradigm, redistributing fixed 74 Gy integral dose in 30 fractions based on mid-treatment FDG PET SUV, generating peaked dose distributions that correspond to SUV values. (D) FLARE-RT composite boost plan with 60 Gy coverage of pre-treatment planning target volume (blue contour), and dose painting-by-numbers boost (rainbow colorwash) to the boost planning target volume (yellow contour), defined on the mid-treatment FDG PET/CT, with dose escalation to >85Gy in areas of highest SUV.

Technical challenges with image-guided adaptive therapy and radiation dose escalation include identifying relevant imaging timepoint(s), defining reproducible radiation targets, generating robust radiation treatment plans, and delivering precise and accurate radiation treatments. The Imaging and Radiation Oncology Core (IROC) under NRG Oncology provides standardized guidelines and supports centralized review of technical data related to imaging and radiation therapy workflows. When incorporating FDG PET/CT into radiation therapy for LA-NSCLC, numerous technical considerations require additional quality assurance steps⁴¹ drawn from the interface of nuclear medicine physics and radiation oncology physics, including guidelines for respiratory-correlated 4D PET/CT and patient-specific motion management during treatment planning and delivery.^{42, 43}

Dose escalation boosts can consist of conventional dose painting to discrete high-risk subvolumes or advanced dose painting-by-numbers with non-uniform prescribed dose at every voxel throughout the target volume based on differential response and risk derived from imaging (FDG PET/CT),⁴⁴ which present tradeoffs with respect to achievable peak planned radiation doses.^{39, 45} While dose painting plans are routinely generated within commercial treatment planning systems, dose painting-by-numbers plans require additional image processing beyond target definition.^{46, 47} One approach for non-uniform voxel-based boost prescription is to discretize volumes into planning contours of increasing dose levels, and optimization objectives are derived from the summary statistics of the voxel dose distribution (i.e. mean, median, upper/lower quartiles, maximum/minimum).^{25, 37} Such objectives shape the boost plan towards a peaked distribution rather than a uniform distribution. Other dose painting-by-numbers inverse planning strategies include the use of an inverted dose distribution as a background, from which uniform total dose is imposed on the optimizer to produce a peaked boost dose distribution.⁴⁸

Whether dose painting or dose painting-by-numbers for response-adaptive dose escalation, there is growing emphasis on generating plans that are robust to inter-fraction and intra-fraction setup uncertainties. Best practice in the current treatment planning paradigm is to use a hybrid set of boost plan optimization objectives that improves the dosimetric solution for the nominal plan, while remaining robust to perturbations under systematic setup and tissue density errors. Emerging paradigms for planning and delivery of increasingly conformal boosts utilizing photon SBRT, pencil beam scanning IMPT, or biological emission-guided radiation therapy (BgRT)⁴⁹ will require daily image guidance with motion management to correct for inter-fraction and intra-fraction setup errors.

Conclusion

The standard of care for patients with locally advanced, unresectable NSCLC is concurrent chemoradiation followed by durvalumab, but efforts to further improve patient outcomes continue. The role of checkpoint inhibitors, next-generation immune modulating agents, and molecularly targeted agents are likely to expand with ongoing trials of combination systemic therapy regimens that intensify treatment over the existing standard of care. Due to unfavorable outcomes following uniform radiation dose escalation and concurrent chemotherapy, radiation trials have since focused on patient selection and personalizing radiation treatment intensification based on tumor biology in addition to tumor anatomy. These advances place increasing demands on precision imaging and treatment planning, to enable personalized radiation plans that could one day integrate clinical, genomic/proteomic/metabolomic, and quantitative radiomic imaging data.