Disparities in Treatment of Patients with Inoperable Stage I Non-Small Cell Lung Cancer: A Population-Based Analysis

Abstract

Background

Patients unable to receive surgery for stage I non-small cell lung cancer (NSCLC) can undergo conventional radiotherapy (ConvRT), stereotactic body radiotherapy (SBRT), or no treatment (NoTx). This study assessed patterns of care and disparities in the receipt of each of these treatments.

Methods

The study included patients in the National Cancer Database from 2003-2011 with T1-T2N0M0 inoperable lung cancer (n= 39,822). Logistic regressions were performed to determine predictors of receiving any radiation vs. NoTx and for receiving SBRT vs. ConvRT.

Findings

Treatment with radiation was significantly less likely in blacks (OR 0.65) and Hispanics (OR 0.42) compared to whites. Treatment with SBRT vs ConvRT was more likely in an academic research program (OR: 2.62) and a high-volume facility (OR: 7.00) compared to community cancer programs or low-volume facilities. In 2011, use of SBRT, ConvRT and NoTx was 25%, 28% and 46% for patients in a community cancer center versus 68%, 11% and 21%, respectively, in an academic center (p < 0.0001).

Conclusion

There were marked institutional and socioeconomic variations in the treatment of inoperable stage I NSCLC. These results suggest that removal of barriers to receive radiation therapy and particularly improved access to SBRT may meaningfully improve survival in this disease.

Introduction

Although surgical resection has been the traditional standard-of-care for curative therapy of stage I non-small cell lung cancer (NSCLC), a significant percentage of patients are inoperable due to comorbid conditions. ¹. This population was historically treated with conventionally fractionated radiotherapy (ConvRT), which necessitated daily treatment over 6-8 weeks, and the survival outcomes were disappointing ^2-5. In fact, the survival following non-surgical treatment was perceived to be so poor that some patients were observed without local therapy (NoTx) under the assumption that any small survival benefit from treatment would be largely mitigated by the patients' competing risks of mortality ^{6, 7}.

Although inoperable stage I lung cancer is not an uncommon entity, there is relatively little information on patterns-of-care in its management. Smith et al. analyzed the treatment of inoperable stage I-II lung cancer in British Columbia, finding that only one-third of patients were treated with curative radiotherapy ⁸ . Similar data are relatively lacking in the United States, which is important not only to appreciate the drivers of an observation approach but also to understand the diffusion and use of novel and improved treatment techniques.

Indeed, stereotactic body radiotherapy (SBRT) is a relatively new treatment modality that delivers precise, high-dose radiotherapy over 1 to 5 fractions ⁹. Both retrospective and prospective studies have shown high local control rates with this technique, results that compare favorably with the surgical literature. For example, RTOG 0236 was a prospective stage II study of SBRT for inoperable stage I lung cancer, and its publication in 2008 revealed a local control probability of 97.6% at 3 years. Previous work has also shown a survival advantage associated with the use of stereotactic body radiotherapy ¹⁰.

In this study, we have used the National Cancer Database to characterize the treatment approaches in patients with inoperable stage I NSCLC diagnosed between 2003 and 2011. Our focus included analyzing the predictors of observation versus active treatment as well as SBRT versus conventional radiotherapy, paying particular notice to treatment trends over time, socioeconomic disparities, and institutional characteristics.

Methods

Database

This study examined the National Cancer Database (NCDB), which is a hospital-based cancer registry that collects data from the American College of Surgeons (ACoS)-Commission on Cancer (CoC) accredited facilities ¹¹. The database is sponsored by the ACoS and the American Cancer Society and includes approximately 70% of all malignant cancers diagnosed in the United States. The database contains information on patient demographics, primary tumor site, histology, site at diagnosis, insurance status, first course of treatment, and overall survival. The NCDB has established specific criteria to ensure the data submitted meets certain quality benchmarks.

Data and Study Population

Eligible patients had histologically confirmed first primary non-small cell lung cancer and received all or part of their first course of treatment at CoC-accredited facilities (if treated at all). In the database, there were 50,603 patients with stage I non-small cell lung cancer treated from 2003-2011 who did not undergo surgery. Patients were stratified into three cohorts: SBRT, ConvRT to a minimum dose of 60 Gy, and no radiation therapy, the latter of whom did not receive any radiation therapy and survived a minimum of 4 months. The 4 month survival criterion was implemented to exclude patients with such short survival that they would presumably been ineligible for any therapy ¹². Patients not meeting criteria for one of these three cohorts were excluded from the analysis, including patients with no information on radiation technique. A total of 39,822 patients met these eligibility criteria.

Statistical Analysis

Candidate variables were grouped into four categories: clinical (e.g. histology, T stage, age, Charlson-Deyo index), time, socioeconomic (e.g. race, insurance status) and institutional (treatment volume, facility type). Facility types were are designated by the Commision on Cancer criteria and include (1) Community Cancer Programs, (2) Comprehensive Community Cancer Programs, and (3) Academic Research Programs. Community cancer programs treat between 100-500 cancer patients a year and have a full range of services for cancer. Comprehensive community cancer programs treat at least 500 cancer patients a year and offer the same range of services. Academic research programs are affiliated with medical schools, have residency programs, and conduct ongoing cancer research. The treatment volume category was calculated by examining the overall total number of patients treated by each facility and dividing it into equal sized tertiles ¹³. Low, medium, and high volume facilities treated a total of 1-12, 13-35, and 36 – 279 patients, respectively. The follow-up time was defined as the time from diagnosis to date of death from any cause or from diagnosis to date of last contact for those who were alive at last contact.

Differences in treatment (NoTx vs. ConvRT vs. SBRT) by clinical, socioeconomic, and institutional characteristics were estimated using the chi-squared test. We also determined predictors of treatment with SBRT vs. ConvRT among patients who received some modality of radiotherapy. Multivariable stepwise logistic regressions were performed to determine independent predictors of any treatment (ConvRT or SBRT) versus no treatment and independent predictors of SBRT among patients treated with radiotherapy. Data were analyzed using SAS version 9.3, JMP version 10 (SAS Institute, Cary, NC) and SPSS v21.

Results

Patient characteristics

A total of 39,822 patients were diagnosed with clinical stage I NSCLC from 2003-2011 and met eligibility criteria for this study. Of this group, 27% (n = 10,687) of patients underwent stereotactic body radiotherapy, 30% (n = 12,163) received daily fractionated conventional radiotherapy, and 43% (n=16,972) underwent no radiation therapy. Patient characteristics are shown in Table 1. Of note, 8.1% of the 18-59 years of age cohort were not insured compared to 0.7% of the rest of the cohort (p < 0.001). Also, throughout the time period there were differences in where patients received treatment with 13.1%, 64.7% and 22.2% being treated at community cancer programs, comprehensive community cancer programs, and academic research programs, respectively, in 2003. This changed to 8.7%, 58.4%, and 32.9% being treated at community cancer programs, comprehensive community cancer programs, and academic research programs, respectively, in 2011 (p < 0.0001).

Table 1. Patient Characteristics among Clinical Stage I Inoperable Non-small Cell Lung Cancer Patients, National Cancer Database 2003-2011.

	All Patients (n=39,822)	No Radiation Therapy (n= 16,972)	Radiation Therapy* (n=22,850)	p - value	Conventional Radiation Therapy (n=12,163)	Stereotactic Body Radiation Therapy (n=10,687)	p - value
Categories
Clinical Factors
Gender				0.895			< 0.0001
Male	49.0%	42.6%	57.4%		55.8%	44.2%
Female	51.0%	42.7%	57.3%		50.8%	49.2%
Comorbidities				0.025			< 0.0001
0	59.7%	43.2%	56.8%		52.0%	48.0%
1	26.2%	41.8%	58.2%		54.9%	45.1%
2	14.1%	41.9%	58.1%		55.2%	44.8%
T stage				< 0.0001			< 0.0001
T1	61.4%	39.7%	60.3%		45.3%	54.7%
T2	38.6%	47.3%	52.7%		67.8%	32.2%
Age at Diagnosis				< 0.0001			< 0.0001
18-59	8.1%	55.5%	44.5%		57.4%	42.6%
60-69	22.4%	42.7%	57.3%		54.6%	45.4%
70-79	39.1%	40.2%	59.8%		53.8%	46.2%
80+	30.4%	42.2%	57.8%		50.7%	49.3%
Histology				< 0.0001			< 0.0001
Adenocarcinoma	42.1%	46.2%	53.8%		46.3%	53.7%
NSCLC (Not Otherwise Specified)	22.2%	41.7%	58.3%		58.5%	41.5%
Squamous	33.4%	38.3%	61.7%		56.8%	43.2%
Large Cell	2.4%	49.0%	51.0%		66.3%	33.8%
Socioeconomic Factors
Race/Ethnicity				< 0.0001			< 0.0001
White	85.8%	40.9%	59.1%		52.7%	47.3%
Black	9.7%	51.9%	48.1%		59.4%	40.6%
Other	2.5%	51.4%	48.6%		51.7%	48.3%
Hispanic	2.0%	62.8%	37.2%		54.2%	45.8%
Insurance Status				< 0.0001			< 0.0001
Not Insured	1.3%	64.8%	35.2%		69.4%	30.6%
Private Insurance/managed Care	15.7%	48.7%	51.3%		54.1%	45.9%
Medicaid	3.6%	52.0%	48.0%		59.2%	40.8%
Medicare	77.6%	41.0%	59.0%		52.7%	47.3%
Other Government	1.8%	22.0%	78.0%		52.1%	47.9%
Institutional Factors
Facility Type				< 0.0001			< 0.0001
Community Cancer Program	11.4%	56.9%	43.1%		84.0%	16.0%
Comprehensive Community Cancer Program	60.4%	43.8%	56.2%		55.9%	44.1%
Academic Research Program	28.1%	34.3%	65.7%		40.2%	59.8%
Facility Volume				< 0.0001			< 0.0001
Low	33.5%	55.3%	44.7%		77.5%	22.5%
Medium	33.2%	43.2%	56.8%		59.7%	40.3%
High	33.3%	29.3%	70.7%		32.6%	67.4%
Year of Diagnosis				< 0.0001			< 0.0001
2003-2005	23.9%	54.6%	45.4%		92.5%	7.5%
2006-2008	32.6%	44.2%	55.8%		61.8%	38.2%
2009-2011	43.5%	34.9%	65.1%		32.7%	67.3%

^*Radiation Treatment cohort includes any patients who received conventional or stereotactic body radiotherapy

Among the cohort, 3% received SBRT from 2003-2005 and 44% received SBRT from 2009-2011. The percentage of patients who did not receive radiation therapy treated from 2003-2005 vs. 2009-2011 dropped from 55% to 35%, and a similar decrease was seen in the percentage of patients treated with conventional radiotherapy, which went from 42% to 21% over that period (Figure 1).

Figure 1. Treatment of Early Stage Inoperable Lung Cancer from 2003-2011.

Predictors of receiving any radiation therapy

The likelihood of undergoing radiation therapy (either SBRT or ConvRT) vs. no radiation therapy is shown in Table 2. Older patients and individuals with T1 stage were more likely to undergo treatment. There was a significant increase in active treatment over the time period of the study. Indicators of less privileged socioeconomic status were strongly related to the probability of receiving treatment: it was less likely in blacks (odds ratio, OR, 0.64) and Hispanics (OR 0.42), and uninsured patients were markedly less likely to undergo treatment. Institutional characteristics also influenced treatment choice, as patients treated at academic and high-volume institutions were significantly more likely to receive treatment. Figure 2 shows the proportion of patients treated with SBRT, ConvRT and No Tx by institution type and year. In 2011, use of SBRT, ConvRT and NoTx was 25%, 28% and 46% for patients in a community cancer center versus 68%, 11% and 21%, respectively, in an academic center (p < 0.0001).

Table 2. Likelihood of Receiving Radiation therapy^* among Clinical Stage I Inoperable Non-small Cell Lung Cancer Patients, National Cancer Database 2003-2011 (n=39,822).

	Odds Ratio	95% (confidence Limits)	p value
Categories
Clinical Factors
Gender
Male	1.0 (ref)		0.75
Female	1.00	0.97 - 1.06
Comorbidities
0	1.0 (ref)		0.50
1	1.03	0.98-1.08
2	1.01	0.95-1.07
T stage
T1	1.0 (ref)		< 0.0001
T2	0.83	0.79-0.86
Age at Diagnosis			< 0.0001
18-59	1.0 (ref)
60-69	1.41	1.29-1.55
70-79	1.49	1.36-1.63
80+	1.39	1.26-1.51
Histology
Adenocarcinoma	1.0 (ref)		< 0.0001
NSCLC (Not Otherwise Specified)	1.42	1.35-1.50
Squamous	1.50	1.43-1.58
Large Cell	1.1	0.96-1.27
Socioeconomic Factors
Race/Ethnicity
White	1.0 (ref)		< 0.0001
Black	0.64	0.60-0.69
Other	0.68	0.59-0.77
Hispanic	0.42	0.36-0.49
Insurance Status
Not Insured	1.0 (ref)		< 0.0001
Private Insurance/managed Care	1.64	1.34-1.99
Medicaid	1.69	1.36-2.09
Medicare	2.02	1.67-2.46
Other Government	4.49	3.44-5.84
Institutional Factors
Facility Type
Community Cancer Program	1.0 (ref)		< 0.0001
Comprehensive Community Cancer Program	1.11	1.03-1.19
Academic Research Program	1.46	1.35-1.59
Facility Volume
Low	1.0 (ref)
Medium	1.48	1.40-1.57	< 0.0001
High	2.55	2.40-2.70
Year of Diagnosis
2003-2005	1.0 (ref)		<0.0001
2006-2008	1.44	1.37-1.53
2009-2011	2.10	1.99-2.21

^*Radiation Treatment cohort includes any patients who received conventional or stereotactic body radiotherapy

Figure 2. Proportion of patients treated with Stereotactic Body Radiotherapy, Conventional Radiation Therapy and No treatment by institution type and year from 2003-2011.

Radiotherapy technique

Predictors of treatment with stereotactic body radiotherapy vs. conventional radiation therapy among patients who received treatment is shown in Table 3. The multivariable analysis echoes the prior model analyzing predictors of any treatment. In particular, black race was significantly associated with a lower probability of receiving SBRT (OR 0.7). Treatment at academic research programs and high treatment volume facilities were associated with a higher likelihood of receiving SBRT (OR 2.67 and 6.6, respectively).

Table 3. Likelihood of Receiving Stereotactic Body Radiotherapy vs. Conventional Radiation Therapy among Clinical Stage I Inoperable Non-small Cell Lung Cancer Patients, National Cancer Database 2003-2011 (n= 22,850).

	Odds Ratio	95% (confidence Limits)	p value
Categories
Clinical Factors
Gender
Male	1.0 (ref)		< 0.0001
Female	1.17	1.09-1.25
Comorbidities
0	1.0 (ref)		< 0.0001
1	0.84	0.78-0.91
2	0.86	0.78-0.94
T stage
T1	1.0 (ref)		< 0.0001
T2	0.46	0.43-0.50
Age at Diagnosis			< 0.0001
18-59	1.0 (ref)
60-69	1.03	0.88-1.20
70-79	1.10	0.95-1.29
80+	1.25	1.07-1.47
Histology
Adenocarcinoma	1.0 (ref)		< 0.0001
NSCLC (Not Otherwise Specified)	0.93	0.85-1.02
Squamous	0.76	0.70-0.82
Large Cell	0.74	0.58-0.94
Socioeconomic Factors
Race/Ethnicity
White	1.0 (ref)		< 0.0001
Black	0.71	0.63-0.80
Other	1.07	0.85-1.34
Hispanic	0.92	0.69-1.22
Insurance Status
Not Insured	1.0 (ref)		0.001
Private Insurance/managed Care	1.99	1.34-2.96
Medicaid	1.65	1.08-2.52
Medicare	2.07	1.40-3.05
Other Government	1.72	1.12-2.67
Institutional Factors
Facility Type
Community Cancer Program	1.0 (ref)		< 0.0001
Comprehensive Community Cancer Program	1.68	1.45-1.95
Academic Research Program	2.67	2.28-3.13
Facility Volume
Low	1.0 (ref)
Medium	1.89	1.72-2.07	< 0.0001
High	6.62	6.02-7.28
Year of Diagnosis
2003-2005	1.0 (ref)		< 0.0001
2006-2008	7.76	6.81-8.84
2009-2011	30.33	26.65-34.51

Discussion

This study has revealed a substantial shift in the patterns of treatment for inoperable stage I non-small cell lung cancer. There has been a dramatic reduction in the number of patients who were observed without active treatment, and a parallel marked increase in the use of SBRT.

Historically, patients who were ineligible to undergo surgical resection would be required to visit a radiation therapy clinic daily for several weeks to receive conventional radiation therapy. Thus one explanation for our results is that the time and effort associated with conventional radiation therapy may have been a significant barrier to receiving treatment, as the falling rates of observation were countered by the rising prevalence of SBRT. A complementary explanation is that the presumptive higher success of SBRT supported the decision for compromised individuals to attempt curative treatment, despite some baseline risk of increasing pulmonary morbidity from treatment. A recent phase II study randomized patients between ConvRT and SBRT. The study found no significant difference in overall survival in patients who received SBRT vs ConvRT (41.3 months vs. 42.1 months, respectively). It concluded that SBRT should be considered as standard treatment for early stage inoperable lung cancer due to less toxicity and patient convenience with less treatment visits and improved cost-effectiveness compared to ConvRT ¹⁴. There are cancer programs where due to financial restrictions or lack of expertise it may be difficult to implement a stereotactic program. In this setting, it may be appropriate to utilize standard 3D conformal radiotherapy techniques with a hypofractionated approach which has also been associated with excellent outcomes and requires less treatment visits compared to the standard conventional approach (30+ treatment visits) ¹⁵.

We found substantial variations in the likelihood of receiving any curative treatment as well as the probability of receiving SBRT among individuals who were irradiated. As shown, a significantly higher proportion of patients were diagnosed and treated at academic research programs in later years which may explain the higher proportion of patients receiving SBRT in later years, as academic centers may have been early adopters of this new technology. Also, black patients and individuals with no insurance were dramatically less likely to be treated for a potentially curable cancer consistent with previous work has shown blacks to less likely receive surgery or chemotherapy compared to whites ^16-18. In patients who did undergo therapy, black patients and individuals with no insurance were also less likely to receive SBRT. Further efforts are needed to better understand and decrease these disparities, whose genesis is likely multifactorial ¹⁹.

Patients who were aged 18-59 years were less likely to receive treatment of SBRT compared to older patients. As medicare provides all Americans above the age > 65 years with health insurance the age disparity could partially be explained by the higher proportion of uninsured patients seen in the younger cohort vs. the rest of the cohort (8.5% vs. 0.7%). Younger age could also be a surrogate for worse access to appropriate medical care which would be independent of insurance status and this could explain why younger patients were less likely to receive any treatment or treatment with SBRT. Health policy initiatives such as the Affordable Care Act in the United States which aims to decrease the number of non-elderly uninsured patients may help alleviate this issue and their impact on reducing cancer disparities should be the subject of future studies.

We also saw significant differences in treatment approach among institutions, as academic centers and high-volume hospitals were significantly more likely to deliver both any treatment and SBRT over conventional RT. Institutional characteristics have previously been shown to influence treatment paradigm for stage IIIA NSCLC and the present findings confirm the importance of institutional characteristics in treatment recommendations ^{20, 21}. These findings may be attributable to these centers having a higher proportion of multi-disciplinary clinics, tumor boards, and navigation programs which all facilitate improved access to care²². It is critical to understand the explanation for these differences and determine whether these differences reflect underlying (and uncontrolled within this analysis) differences in patient demographics, treatment recommendation biases, and/or technology gaps.

This study does have limitations implicit in large retrospective database analyses including institution reporting bias. Also, the National Cancer Database does not record several important variables, such as Karnofsky performance status, weight loss, or tumor location (central vs. peripheral) which may have affected the treatment received. Another limitation of the study is that survival information was unavailable for the entire time period in this study and therefore these results cannot be linked to survival outcomes.

In conclusion, we have shown variable diffusion of a new treatment technique (SBRT) among the general population and we should consider mechanisms to improve access to SBRT ²³. For example, all patients with early-stage lung cancer who are ineligible to undergo surgery should be evaluated by a radiation oncologist prior to excluding curative radiotherapy. Similarly, there should be a focus on education and training of radiation oncologists in stereotactic techniques, improving access to facilities that offer SBRT, and cancer accreditation bodies requiring facilities to offer SBRT as a treatment modality. Based on our data, such relatively straightforward efforts may lead to meaningful improvements in survival in this challenging patient demographic.