Background

Surgical therapy is the mainstay of treatment of early-stage non–small cell lung cancer (NSCLC), particularly in the node-negative setting. For medically inoperable patients with node-negative disease, curative therapy typically involves definitive radiotherapy (RT). Although stereotactic body RT (SBRT) alone has become a standard therapy for small inoperable tumors, the optimal treatment of patients with larger (5–7 cm) tumors is less clear.

The role of chemotherapy for early-stage NSCLC is best defined in the postoperative setting and is frequently used for tumors >4 cm or when other high-risk features are present.^1–6 For patients with inoperable node-negative disease receiving definitive RT, however, prospective data evaluating the role of chemotherapy are lacking. As a result, based on extrapolations of surgical data, NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines) for NSCLC recommend the consideration of chemotherapy for high-risk patients with inoperable disease.⁷ In addition to the uncertainties surrounding chemotherapy use, data are also lacking regarding ablative SBRT techniques for larger tumors, given that patients with tumors >5 cm were generally excluded from major prospective SBRT trials.^8–12

In this study, we sought to analyze practice patterns for patients with large (5–7 cm), node-negative NSCLC with regard to chemotherapy use and RT technique. Given the lack of prospective data defining optimal treatment, we hypothesized that treatment paradigms would commonly stray from NCCN Guidelines and sought to identify factors associated with receipt of NCCN-nonadherent therapy. The impact of treatment paradigm on overall survival (OS) was also evaluated.

Patients and Methods

Results

Based on inclusion and exclusion criteria, a total of 2,020 patients with 5- to 7-cm node-negative NSCLC were included in the final cohort (supplemental eFigure 1). Median follow-up was 18.7 months. Median age was 75 years, with most patients being male (57.9%), non-Hispanic White (81.8%), and having a CD comorbidity score of 0 to 1 (80.3%). Other demographic factors for the total cohort are presented in Table 1. Tumor size was between 5 and 6 cm for most patients (61.8%), and the remainder of patients had tumor sizes between 6 and 7 cm. Nearly half of the patients (49.3%) received chemotherapy as part of their initial, definitive treatment, with 32.2% receiving concurrent chemoRT and 17.1% receiving sequential RT and chemotherapy. Most patients received CFRT (64.3%), whereas SBRT (21.8%) and HFRT (13.9%) were less common. The most commonly used RT fractionation regimen for each category can be found in supplemental eTable 1.

Table 1.

Patient Characteristics

View Table

Figure 1 summarizes the initial treatment of patients with regard to chemotherapy use and RT type. The most common treatment paradigm was concurrent chemoRT using conventional fractionation, with 31% of patients receiving this treatment despite the lack of support by NCCN Guidelines.⁷ Patients receiving SBRT (22%) and HFRT (14%) only rarely received chemotherapy, with most being treated with definitive RT alone.

View Full Size

Patterns of chemotherapy and radiotherapy use.

Abbreviations: CFRT, conventionally fractionated radiotherapy; HFRT, hypofractionated radiotherapy; SBRT, stereotactic body radiotherapy.

Citation: Journal of the National Comprehensive Cancer Network 20, 4; 10.6004/jnccn.2021.7043

Given the high rate of nonadherence to NCCN Guidelines regarding the use of concurrent chemoRT, we investigated factors associated with receipt of NCCN-nonadherent therapy. Patients receiving concurrent chemoRT were classified as NCCN-nonadherent, whereas all other patients (ie, those receiving definitive RT alone or RT with sequential chemotherapy) were classified as NCCN-adherent. Clinicopathologic and demographic factors of these 2 patient populations are reported in supplemental eTable 2. To determine factors associated with selection of NCCN-nonadherent therapy, univariable and multivariable regressions were performed (Table 2). In univariable analysis, younger age (<70 years), male sex, private insurance, treatment at a nonacademic center, and larger tumor size (≥6.0 cm) were associated with receipt of NCCN-nonadherent therapy. In multivariable analysis, age <70 years (OR, 2.72; P<.001), treatment at a nonacademic center (OR, 1.65; P<.001), and larger tumor size (≥6.0 cm; OR, 1.27; P=.026) remained statistically associated with receipt of NCCN-nonadherent therapy.

Table 2.

Factors Associated With Nonadherence to NCCN Guidelines

View Table

We evaluated whether the use of NCCN-nonadherent concurrent chemoRT changed during the study period (Figure 2). There was a statistically similar number of patients receiving NCCN-nonadherent chemoRT from 2004 to 2016 (range, 28%–40%; Spearman ρ = 0.02; P=.415). With regard to RT type, practice patterns changed substantially with increasing use of SBRT (Spearman ρ = 0.18; P<.0001) and, to a lesser extent, HFRT (Spearman ρ = 0.06; P=.098). In particular, SBRT use jumped from the 0% to 10% range in 2004 through 2007 to 25% to 30% by 2013 through 2016. This was accompanied by a corresponding decreased use of conventional fractionation over the same time period. Patient factors associated with the use of SBRT were analyzed with univariable and multivariable logistic regressions (supplemental eTable 3). In univariable analysis, older age (>70 years), non-Hispanic White race, year diagnosed (2011–2016), nonprivate insurance, treatment at an academic center, and smaller tumor size (<6.0 cm) were significantly associated with receipt of SBRT. In subsequent multivariable analysis, age ≥70 years (OR, 1.64; P<.001), non-Hispanic White race (OR, 1.57; P=.011), year of diagnosis 2011 through 2016 (OR, 2.48; P<.001), treatment at an academic center (OR, 3.24; P<.001), and tumor size <6 cm (OR, 2.52; P<.001) were statistically associated with receipt of SBRT.

View Full Size

Trends in (A) chemotherapy and (B) radiotherapy use over time in our cohort. A statistically significant trend of increased use of SBRT (Spearman ρ = 0.1775; P<.001) and HFRT was found (Spearman ρ = 0.058; P=.010) with corresponding decreased use of CFRT (Spearman ρ = −0.1955; P<.001). Comparison of (C) chemotherapy and (D) radiotherapy use from 2004 through 2010 and 2011 through 2016.

Abbreviations: CFRT, conventionally fractionated radiotherapy; HFRT, hypofractionated radiotherapy; SBRT, stereotactic body radiotherapy.

^*P<.001 by chi-square analysis.

Citation: Journal of the National Comprehensive Cancer Network 20, 4; 10.6004/jnccn.2021.7043

The effect of treatment pattern on OS was evaluated. Supplemental eFigure 2 illustrates Kaplan-Meier survival curves for patients based on the use of chemotherapy and RT type. In log-rank analysis, patients receiving either sequential or concurrent chemotherapy had prolonged survival compared with those not receiving chemotherapy (P<.001). However, there was no difference between patients receiving sequential chemotherapy and those receiving concurrent chemotherapy (P=.895). Median survival data and log-rank analyses are reported for each treatment group in supplemental eTables 4 and 5. For the entire cohort, median OS was 21 months (95% CI, 19.8–22.1). The longest median OS was observed for patients receiving SBRT with sequential chemotherapy (36 months; 95% CI, 25.7–53.4), although this was not statistically significant compared with all other groups receiving chemotherapy.

To examine factors associated with survival, and in particular to assess the impact of NCCN-adherent versus NCCN-nonadherent therapy, univariable and multivariable regression analyses were performed (Table 3). Although NCCN-adherent versus NCCN-nonadherent therapy was associated with increased risk of mortality in univariable analysis, multivariable analysis failed to show a significant effect when controlling for other factors (hazard ratio [HR], 1.00; 95% CI, 0.85–1.18; P=.992). Factors that were associated with increased risk of mortality included older age (≥70 vs <70 years; HR, 1.17; 95% CI, 1.04–1.33; P=.011), male versus female sex (HR, 1.28; 95% CI, 1.15–1.43; P<.001), CD score ≥2 versus 0 to 1 (HR, 1.25; 95% CI, 1.09–1.43; P=.001), diagnosis year 2004 through 2010 versus 2011 through 2016 (HR, 1.15; 95% CI, 1.03–1.28; P=.014), nonprivate insurance versus private insurance (HR, 1.27; 95% CI, 1.08–1.49; P=.004), tumor size ≥6.0 versus <6.0 cm (HR, 1.21; 95% CI, 1.09–1.35; P=.001), and no chemotherapy versus chemotherapy (HR, 1.27; 95% CI, 1.09–1.49; P=.002).

Table 3.

Factors Associated With Mortality

View Table

Discussion

The optimal treatment of patients with inoperable, large, node-negative NSCLC remains unclear. NCCN Guidelines⁷ recommend definitive RT with or without adjuvant chemotherapy for these patients, based mostly on extrapolations from similarly staged surgical patients, in whom adjuvant chemotherapy improves outcomes for patients with high-risk features (eg, tumors >4 cm).^1–6 Concurrent chemoRT is not currently recommended by NCCN Guidelines,⁷ because key trials defining the role of definitive chemoRT in locally advanced NSCLC generally excluded node-negative patients.^15–19

In this study, we found that roughly one-third of patients with NSCLC with 5- to 7-cm primary tumors received concurrent chemoRT, which is not supported by NCCN Guidelines.⁷ Other treatment strategies, including SBRT, HFRT, or CFRT with or without sequential chemotherapy, were used less often than NCCN-nonadherent chemoRT. Multivariable analysis suggests that clinicians elected the concurrent chemoRT treatment strategy for younger (presumably more fit) patients with larger tumors. Interestingly, the use of concurrent chemoRT remained stable over the entire time period studied (2004–2016), suggesting that this is an engrained practice among some physicians, regardless of current national guidelines.

Although the use of concurrent chemoRT in node-negative patients is not backed by high-level data, the frequent use of this approach is not necessarily surprising. It is known that patients with larger primary tumors are at high risk for regional nodal and distant failure.^20,21 Specifically among patients with primary tumors ≥5 cm, retrospective SBRT series have shown distant failure to be the most common mode of failure, with 31% to 33% of all patients developing distant metastases.^22–24 Data from several prospective studies have similarly shown high rates of distant failure for node-negative NSCLC, with RTOG 0236 reporting a distant failure rate of 45% at 5 years for patients with T2 tumors⁸ and RTOG 0917 reporting rates of any distant failure to be 44% to 47%.¹⁰ In locally advanced NSCLC, concurrent chemoRT is known to be superior in terms of locoregional control and OS compared with sequential RT in several randomized controlled trials and meta-analyses.^15–17,25 Although increased esophageal toxicity is expected among node-positive patients receiving RT to the mediastinum, there may be a lower perceived risk of toxicity in our study cohort of node-negative patients, in whom lower RT doses to the esophagus are expected.

We did not observe an association between OS and adherence to NCCN Guidelines when adjusting for other factors in multivariable analysis. Although a survival benefit with receipt of NCCN-nonadherent therapy was seen in univariable analysis, this was likely a result of the confounding variable of chemotherapy, which by definition was administered to all patients receiving NCCN-nonadherent concurrent chemoRT. In contrast, only 25% of patients in the NCCN-nonadherent group received chemotherapy. In multivariable survival analysis, lack of receipt of chemotherapy was associated with worse OS. Although this could partially be attributed to selection bias (with fitter patients receiving chemotherapy), studies in similar cohorts of patients receiving definitive RT have consistently shown an OS benefit with the addition of chemotherapy^13,26,27 to definitive RT. Despite this, only half of our cohort received chemotherapy, likely due to comorbidities and poor functional status that are common in patients with NSCLC (eg, older age, smoking history, chronic obstructive pulmonary disease). For patients unable to tolerate traditional chemotherapy, immunotherapy is an emerging potential systemic treatment option that may reduce distant failure and improve outcomes in this population,²⁸ with several trials underway investigating SBRT + immunotherapy in patients with T1–3N0M0 NSCLC (eg, ClinicalTrials.gov identifiers: NCT03110978 and NCT03833154). Although the NCDB did begin to include an immunotherapy variable starting in 2013, this variable does not include standard immunotherapy agents used in NSCLC (eg, durvalumab and other PD-1/PD-L1 inhibitors); thus, it remains unknown whether patients in our cohort received immunotherapy .

The classification of SBRT in this study was based on US billing standards, in which SBRT is defined as ≤5 fractions. However, other more protracted regimens with high biologically effective doses (BEDs; BED at α/β ratio of 10 [BED₁₀], >100) are often also considered SBRT. For example, 2 regimens (7 Gy × 10 fractions = 70 Gy and 7.5 Gy × 8 fractions = 60 Gy) were used in 33 patients in our study, who were included in our HFRT cohort. We performed a sensitivity analysis using a more inclusive SBRT definition (BED₁₀, >100) and found that the major results and conclusions of this study were unchanged, regardless of classification of these patients (data not shown). Similarly, for the classification of chemotherapy as sequential or concurrent, we performed a sensitivity analysis assessing the effect of defining concurrent chemotherapy as 3 days, 10 days, or 14 days within the initiation of RT. No substantial changes in our main conclusions were seen (data not shown).

This study has several limitations. First, as previously mentioned, given the retrospective nature of this study, there is inherent selection bias that may have influenced treatment selection and therefore survival analyses. Although multivariable analyses were performed to try to account for some of these factors, there are other factors that may have affected treatment selection that were not incorporated, including some that are not available in the NCDB, such as patient performance status and medical contraindications to chemotherapy. Second, although the goal of this study was to characterize patterns of care in the United States for patients with large, node-negative NSCLC treated with definitive RT, the NCDB is not a comprehensive population-based dataset, and it includes data from CoC member institutions only. Although the NCDB does capture >70% of new cancer diagnoses in the United States, if practice patterns varied greatly between CoC member and non-CoC member sites, the overall described practice patterns may not perfectly reflect national trends.

Conclusions

Practice patterns vary greatly among patients with 5- to 7-cm node-negative NSCLC receiving definitive RT. A substantial proportion of patients were not treated in concordance with NCCN Guidelines (Version 3.2020) and received concurrent chemoRT. The proportion of patients receiving NCCN-nonadherent concurrent chemoRT did not vary over time. Although younger patients with large tumors treated at nonacademic medical centers were more likely to receive NCCN-nonadherent therapy, adherence to NCCN Guidelines was not associated with differences in OS.