Abstract
Purpose:
Concerns have been raised about the potential for worse treatment outcomes because of dosimetric inaccuracies related to tumor motion and increased toxicity caused by the spread of low-dose radiation to normal tissues in patients with locally advanced non-small cell lung cancer (NSCLC) treated with intensity modulated radiation therapy (IMRT). We therefore performed a population-based comparative effectiveness analysis of IMRT, conventional 3-dimensional conformal radiation therapy (3D-CRT), and 2-dimensional radiation therapy (2D-RT) in stage III NSCLC.
Methods and Materials:
We used the Surveillance, Epidemiology, and End Results (SEER)-Medicare database to identify a cohort of patients diagnosed with stage III NSCLC from 2002 to 2009 treated with IMRT, 3D-CRT, or 2D-RT. Using Cox regression and propensity score matching, we compared survival and toxicities of these treatments.
Results:
The proportion of patients treated with IMRT increased from 2% in 2002 to 25% in 2009, and the use of 2D-RT decreased from 32% to 3%. In univariate analysis, IMRT was associated with improved overall survival (OS) (hazard ratio [HR] 0.90, P = .02) and cancer-specific survival (CSS) (HR 0.89, P = .02). After controlling for confounders, IMRT was associated with similar OS (HR 0.94, P = .23) and CSS (HR 0.94, P = .28) compared with 3D-CRT. Both techniques had superior OS compared with 2D-RT. IMRT was associated with similar toxicity risks on multivariate analysis compared with 3D-CRT. Propensity score matched model results were similar to those from adjusted models.
Conclusions:
In this population-based analysis, IMRT for stage III NSCLC was associated with similar OS and CSS and maintained similar toxicity risks compared with 3D-CRT.
Summary
We used Surveillance, Epidemiology, and End Results-Medicare data to perform a comparative effectiveness analysis of radiation therapy techniques for stage III NSCLC. From 2002 to 2009, the use of 2-dimensional radiation therapy declined, and the adoption of 3-dimensional conformal radiation therapy (3D-CRT) and intensity modulated radiation therapy (IMRT) increased. Multivariate adjusted and propensity score matching analysis demonstrated similar overall and cancer-specific survival and similar toxicity profile with IMRT compared with 3D-CRT.
Introduction
The standard of care for stage III non-small cell lung cancer (NSCLC) includes a combination of radiation therapy (RT), chemotherapy, and potentially surgery in well-selected patients (1). Patients treated with thoracic RT are at risk for a variety of toxicities, including upper gastrointestinal (UGI), pulmonary, and cardiac. Historically, 2-dimensional simulation and planning (2D-RT) with a limited number of beams were used to treat locally advanced NSCLC. Improvements in simulation, planning, and delivery technology led to the development of 3-dimensional conformal radiation therapy (3D-CRT). More recently, intensity modulated radiation therapy (IMRT) has been used in an attempt to further decrease the volume of normal tissue exposed to high doses of radiation (2, 3).
Although the ability to spare organs at risk makes use of IMRT attractive, it also raises several potential concerns. First, IMRT increases the amount of normal lung tissue exposed to low doses of radiation and could potentially increase the risk of pneumonitis. Second, inasmuch as lung tumors move with breathing, interplay between motion of the tumor and the multileaf collimator (MLC)-shaped segments could result in unanticipated variation in the dose delivered to the target (4, 5). Ultimately, these theoretical concerns are best addressed by comparing outcomes and toxicities of IMRT with those of other radiation treatment methods in locally advanced NSCLC. Studies addressing this question have been limited to single-institution retrospective comparisons and single-arm studies (6–9), and no randomized-controlled trials have been completed.
We performed a population-based comparative effectiveness analysis of radiation treatment strategies for locally advanced NSCLC. We used the Surveillance, Epidemiology, and End Results (SEER)-Medicare database to compare IMRT, 3D-CRT, and 2D-RT with respect to patient survival and toxicity in a large cohort of elderly patients diagnosed with stage III NSCLC.
Methods and Materials
Data source
We analyzed the National Cancer Institute’s SEER-Medicare linked dataset. SEER represents information gathered over 17 geographic areas that account for 26% of the United States population (10). The linked dataset contains demographic, clinical, pathologic, and Medicare insurance claim data. Patients identified from SEER from January 1, 2002, through December 31, 2009, were linked to Medicare claims through December 31, 2010.
Cohort construction
During 2002 to 2009, 183,965 patients were diagnosed with pathologically confirmed NSCLC. Our cohort included patients aged ≥65 with stage III NSCLC. Patients were enrolled in Medicare parts A and B for 12 months before diagnosis until death or censoring, and they were excluded for enrollment in a health maintenance organization, a diagnosis at death, or an invalid diagnosis date.
We excluded patients who according to SEER did not receive RT or who were diagnosed with a malignant pleural effusion. Staging was according to the third edition of the American Joint Committee on Cancer (11). This study was granted exempt status by our institutional review board.
Identification of RT use
Radiation therapy was identified by Healthcare Common Procedure Coding System (HCPCS) codes for radiation within 6 months from diagnosis. RT course length was the number of days between the first and last RT-related claims, and patients were excluded for a course length of <3 weeks or >9 weeks. IMRT was identified by the presence of at least 1 Medicare claim for IMRT treatment or plan. Additionally, 3D-CRT was identified by the presence of claims for the use of either complex simulation with computed tomographic guidance or 3D simulation. Further, 2D-RT was identified by the presence of claims for simple or intermediate simulation. Table E1 (available online at www.redjournal.org) contains detailed coding information.
Baseline characteristics
Demographic information from SEER included age, sex, race, marital status, geographic area (West, Midwest, South, Northeast), urban setting, area educational attainment (≥4 years of college), and area median income.
Disease characteristics from SEER data included tumor site and laterality, histology, grade, tumor size, and nodal status. A modified Charlson-Deyo comorbidity index and cardiac risk factors were calculated from Medicare claims 12 months before diagnosis (12, 13), but chronic obstructive pulmonary disease was considered a separate variable. Diabetes was also excluded from the comorbidity score and included as a cardiac risk factor, along with hypertension, hyperlipidemia/hypercholesterolemia, and atherosclerosis. Oxygen use was determined from claims for home oxygen supplies. A performance score was calculated using claims indicating hospitalization; skilled nursing or long-term care facility stay; home health agency use; and claims for walkers, canes, crutches, wheelchairs, diabetic footwear, commodes, or hospital beds. This score is related to poor health status (14).
Diagnosis and treatment
Pretreatment positron emission tomography (PET) and brain imaging, which included magnetic resonance imaging and computed tomography, were determined from claims 3 months before RT. Invasive mediastinal staging was determined from claims 3 months before diagnosis until RT and included video-assisted thoracic mediastinal biopsy, bronchoscopy with accompanying claim for nodal biopsy, and mediastinoscopy/mediastinotomy.
Surgery and chemotherapy were assessed in the 6 months after diagnosis and were determined from Medicare and SEER data. Surgery included sublobar resection, lobar resection, pneumonectomy, and unclassified lung resection. Chemotherapy was stratified by concurrent or sequential and included carboplatin-paclitaxel, cisplatin-etoposide, or other agent(s). Doublets were identified by claims for the second agent within 8 days of the first, similar to previous studies (15, 16). Facility was categorized as a freestanding center, hospital-based National Cancer Institute (NCI) center, or hospital-based non-NCI center.
Endpoint classification
Overall survival (OS) and cancer-specific survival (CSS) were determined from Medicare and SEER records. Composite UGI, pulmonary, and cardiac toxicity variables were created using ICD-9-CM diagnostic and procedural codes, diagnosis-related groups, and HCPCS codes (Table E1). UGI toxicity was categorized as early or late using a 4-week from treatment cutoff, which captures most events (9, 17). Early UGI toxicity included esophagitis, mucositis, gastroenteritis, esophageal ulcer, dehydration, dysphagia, and tube feeding. Late UGI toxicity included esophageal stricture or repair of stricture. Pulmonary toxicity was categorized as early or late using a cutoff 6 months from treatment. Early pulmonary toxicity included intubation and unspecified acute pulmonary toxicity resulting from RT. Late pulmonary toxicity included pulmonary fibrosis and unspecified chronic pulmonary toxicity resulting from RT. Cardiac toxicity was examined as a late complication of RT (>6 months) and included coronary artery disease/atherosclerosis, pericarditis or pericardial effusion, pericardiocentesis, myocardial infarct or ischemia, revascularization procedure, heart failure, and conduction disorder/dysrhythmia. To reduce misclassification of comorbidities, cardiac events up to 6 months after RT were considered preexistent.
Statistical analysis
Baseline characteristics were compared between groups using χ2 tests. Kaplan-Meier methodology was used to estimate outcomes, with overall comparisons accomplished using the log–rank statistic. Time was measured from the date of diagnosis or first radiation treatment for survival and toxicity endpoints, respectively. Aside from the OS analysis, patients were censored at death or at the last follow-up visit. Cox proportional hazards models were used to control for confounders, on the basis of model building criteria P < .2 to enter and remain in the model. All models were additionally adjusted for the following variables: age, year of diagnosis, comorbidity index, chronic obstructive pulmonary disease, oxygen use, performance score, cardiac risk factors, tumor histology, tumor grade, tumor size, nodal stage, PET, brain imaging, invasive mediastinal evaluation, surgery, chemotherapy, facility type, and RT course length. To evaluate the proportional hazards assumption, log-log plots were examined visually, and time-interaction effects were tested for statistical significance. Goodness of fit was assessed with the likelihood ratio.
We used a more robust propensity score matching analysis to compare IMRT with 3D-CRT. Propensity scores were calculated using a multivariable logistic regression with IMRT regressed on all available variables. We used the nearest neighbor, 3:1 variable matching approach with a maximum caliper of 0.2 times the standard deviation of the logit of the propensity score (18). Balance was assessed with a 20% maximum standardized difference (19).
Statistical significance was set at .05, and all tests relied on 2-tailed P values. No adjustments were made to account for multiplicity. All statistical analyses were performed using SAS (version 9.3, SAS, Cary, NC).
Results
Using the SEER-Medicare database, we identified a cohort of 6894 patients diagnosed with stage IIIA/B NSCLC between 2002 and 2009 (Fig. E1, available online at www.redjournal.org). The use of IMRT increased from 2% in 2002 to 25% in 2009, and the use of 2D-RT decreased from 32% to 3% (P for trend <.0001, Fig. 1A). The median age was 74, and 45% of patients were female. Most patients received chemotherapy (69%), with concurrent carboplatin-paclitaxel (20%) being more common than concurrent cisplatin-etoposide (3%). Baseline characteristics are shown in Table 1.
Fig. 1.
Prevalence and univariate analysis of 2-dimensional radiation therapy (2D-RT), 3-dimensional conformal RT (3D-CRT), and intensity modulated RT (IMRT) for stage III non-small cell lung cancer. (A) RT technique by year of diagnosis. Kaplan-Meier analysis of IMRT, 3D-CRT, and 2D-RT. (B) 2D-RT results in inferior overall survival on proportional hazards models (P <.0001), whereas IMRT is superior to 3D-CRT (hazard ratio [HR] 0.90, P = .02). (C) 2D-RT results in inferior cancer-specific survival (P <.0001), and IMRT is superior to 3D-CRT (HR 0.89, P = .02).
Table 1.
Demographic and clinical characteristics of patients with Stage III NSCLC treated with radiation
| Characteristic | Overall cohort n = 6894 (%) | IMRT n = 716 (%) | 3D-CRT n = 5356 (%) | 2D-RT n = 822 (%) | P (IMRT vs 3D) | P (all groups) |
|---|---|---|---|---|---|---|
|
| ||||||
| Age | ||||||
| 65–69 | 1605 (23) | 164 (23) | 1245 (23) | 196 (24) | .75 | .66 |
| 70–74 | 2048 (30) | 223 (31) | 1564 (29) | 261 (32) | ||
| 75–79 | 1763 (26) | 181 (25) | 1386 (26) | 196 (24) | ||
| ≥80 | 1478 (21) | 148 (21) | 1161 (22) | 169 (21) | ||
| Sex | ||||||
| Male | 3799 (55) | 377 (53) | 2962 (55) | 460 (56) | .18 | .36 |
| Female | 3095 (45) | 339 (47) | 2394 (45) | 362 (44) | ||
| Race | ||||||
| White | 5788 (84) | 584 (82) | 4564 (85) | 640 (78) | .01 | <.0001 |
| Black | 605 (9) | 67 (9) | 462 (9) | 76 (9) | ||
| Hispanic | 213 (3) | 33 (5) | 149 (3) | 31 (4) | ||
| Other | 288 (4) | 32 (4) | 181 (3) | 75 (9) | ||
| Marital status | ||||||
| Unmarried | 2963 (43) | 302 (42) | 2296 (43) | 365 (44) | .93 | .57 |
| Married | 3780 (55) | 398 (56) | 2937 (55) | 445 (54) | ||
| Unknown | 151 (2) | 16 (2) | 123 (2) | 12 (1) | ||
| Geographic area | ||||||
| West | 2355 (34) | 275 (38) | 1769 (33) | 311 (38) | .004 | .0004 |
| Midwest | 1103 (16) | 94 (13) | 900 (17) | 109 (13) | ||
| South | 1878 (27) | 201 (28) | 1441 (27) | 236 (29) | ||
| Northeast | 1558 (23) | 146 (20) | 1246 (23) | 166 (20) | ||
| Urban setting | ||||||
| Urban | 6124 (89) | 664 (93) | 4772 (89) | 688 (84) | .003 | <.0001 |
| Rural | 770 (11) | 52 (7) | 584 (11) | 134 (16) | ||
| Year of diagnosis | ||||||
| 2002 | 774 (11) | 16 (2) | 512 (10) | 246 (30) | <.0001 | <.0001 |
| 2003 | 849 (12) | 21 (3) | 653 (12) | 175 (21) | ||
| 2004 | 826 (12) | 27 (4) | 669 (12) | 130 (16) | ||
| 2005 | 783 (11) | 48 (7) | 656 (12) | 79 (10) | ||
| 2006 | 785 (11) | 69 (10) | 644 (12) | 72 (9) | ||
| 2007 | 991 (14) | 122 (17) | 817 (15) | 52 (6) | ||
| 2008 | 978 (14) | 189 (26) | 750 (14) | 39 (5) | ||
| 2009 | 908 (13) | 224 (31) | 655 (12) | 29 (4) | ||
| Educational attainment of census tract or zip code | ||||||
| Quartile 1 | 1685 (24) | 170 (24) | 1270 (24) | 245 (30) | .77 | .01 |
| Quartile 2 | 1724 (25) | 173 (24) | 1359 (25) | 192 (23) | ||
| Quartile 3 | 1754 (25) | 180 (25) | 1374 (26) | 200 (24) | ||
| Quartile 4 | 1731 (25) | 193 (27) | 1353 (25) | 185 (23) | ||
| Median income of census tract or zip code | ||||||
| Quartile 1 | 1680 (24) | 171 (24) | 1285 (24) | 224 (27) | .45 | .31 |
| Quartile 2 | 1734 (25) | 166 (23) | 1363 (25) | 205 (25) | ||
| Quartile 3 | 1736 (25) | 182 (25) | 1359 (25) | 195 (24) | ||
| Quartile 4 | 1744 (25) | 197 (28) | 1349 (25) | 198 (24) | ||
| Comorbidity | ||||||
| 0 | 5041 (73) | 521 (73) | 3923 (73) | 597 (73) | .03 | .11 |
| 1 | 1157 (17) | 104 (15) | 907 (17) | 146 (18) | ||
| 2 | 430 (6) | 52 (7) | 333 (6) | 45 (5) | ||
| ≥3 | 266 (4) | 39 (5) | 193 (4) | 34 (4) | ||
| COPD status | ||||||
| No | 4371 (63) | 452 (63) | 3403 (64) | 516 (63) | .83 | .90 |
| Yes | 2523 (37) | 264 (37) | 1953 (36) | 306 (37) | ||
| Oxygen-dependent status | ||||||
| No | 6297 (91) | 633 (88) | 4908 (92) | 756 (92) | .004 | .01 |
| Yes | 597 (9) | 83 (12) | 448 (8) | 66 (8) | ||
| Cardiovascular risk factors | ||||||
| 0 | 1745 (25) | 140 (20) | 1359 (25) | 246 (30) | <.0001 | <.0001 |
| 1 | 2086 (30) | 204 (28) | 1617 (30) | 265 (32) | ||
| 2 | 2083 (30) | 236 (33) | 1636 (31) | 211 (26) | ||
| ≥3 | 980 (14) | 136 (19) | 744 (14) | 100 (12) | ||
| Performance status | ||||||
| 0 | 6105 (89) | 617 (86) | 4775 (89) | 713 (87) | .06 | .05 |
| 1 | 521 (8) | 66 (9) | 387 (7) | 68 (8) | ||
| ≥2 | 268 (4) | 33 (5) | 194 (4) | 41 (5) | ||
| Tumor site | ||||||
| Upper lobe | 4028 (58) | 422 (59) | 3148 (59) | 458 (56) | .62 | <.0001 |
| Middle lobe | 254 (4) | 24 (3) | 185 (3) | 45 (5) | ||
| Lower lobe | 1684 (24) | 188 (26) | 1326 (25) | 170 (21) | ||
| Other | 928 (13) | 82 (11) | 697 (13) | 149 (18) | ||
| Tumor laterality | ||||||
| Left | 2857 (41) | 314 (44) | 2204 (41) | 339 (41) | .29 | .53 |
| Right | >3951 (>57) | >391 (>54) | 3087 (58) | >472 (>57) | ||
| Other/both | <86 (<2) | <11 (<2) | 65 (1) | <11 (<2) | ||
| Histology | ||||||
| Adenocarcinoma | 1974 (29) | 214 (30) | 1545 (29) | 215 (26) | .58 | .42 |
| Squamous | 2795 (41) | 294 (41) | 2154 (40) | 347 (42) | ||
| Other NSCLC | 2125 (31) | 208 (29) | 1657 (31) | 260 (32) | ||
| Grade | ||||||
| Low | 161 (2) | 15 (2) | 132 (2) | 14 (2) | .42 | .0001 |
| Intermediate | 1201 (17) | 113 (16) | 951 (18) | 137 (17) | ||
| High | 2344 (34) | 243 (34) | 1784 (33) | 317 (39) | ||
| Undifferentiated | 186 (3) | 13 (2) | 134 (3) | 39 (5) | ||
| Unknown | 3002 (44) | 332 (46) | 2355 (44) | 315 (38) | ||
| Tumor size | ||||||
| ≤2 cm | 585 (8) | 66 (9) | 460 (9) | 59 (7) | .95 | .0001 |
| 2.1–3 cm | 844 (12) | 89 (12) | 659 (12) | 96 (12) | ||
| 3.1–5 cm | 1997 (29) | 219 (31) | 1574 (29) | 204 (25) | ||
| 5.1–7 cm | 1349 (20) | 137 (19) | 1059 (20) | 153 (19) | ||
| >7 cm | 924 (13) | 92 (13) | 721 (13) | 111 (14) | ||
| Unknown | 1195 (17) | 113 (16) | 883 (16) | 199 (24) | ||
| Nodal status | ||||||
| N0 | 1251 (18) | 132 (18) | 950 (18) | 169 (21) | .02 | .001 |
| N1 | 388 (6) | 35 (5) | 314 (6) | 39 (5) | ||
| N2 | 4227 (61) | 433 (60) | 3308 (62) | 486 (59) | ||
| N3 | >910 (>13) | >105 (>14) | 694 (13) | 102 (12) | ||
| Unknown | <127 (<2) | <11 (<2) | 90 (2) | 26 (3) | ||
| PET scan | ||||||
| Not performed | 4625 (67) | 271 (38) | 3618 (68) | 736 (90) | <.0001 | <.0001 |
| Performed | 2269 (33) | 445 (62) | 1738 (32) | 86 (10) | ||
| Brain imaging | ||||||
| Not performed | 3111 (45) | 275 (38) | 2422 (45) | 414 (50) | .001 | <.0001 |
| Performed | 3783 (55) | 441 (62) | 2934 (55) | 408 (50) | ||
| Invasive mediastinal staging | ||||||
| Not examined | 5633 (82) | 583 (81) | 4347 (81) | 703 (86) | .87 | .01 |
| Examined | 1261 (18) | 133 (19) | 1009 (19) | 119 (14) | ||
| Surgery | ||||||
| None | 5854 (85) | 624 (87) | 4561 (85) | 669 (81) | .04 | .003 |
| Sublobectomy | 198 (3) | 25 (3) | 140 (3) | 33 (4) | ||
| Lobectomy/pneumonectomy | 842 (12) | 67 (9) | 655 (12) | 120 (15) | ||
| Chemotherapy | ||||||
| None | 2105 (31) | 183 (26) | 1567 (29) | 355 (43) | .09 | <.0001 |
| Carboplatin-paclitaxel | ||||||
| Concurrent | 1382 (20) | 146 (20) | 1115 (21) | |||
| Sequential | 195 (3) | 27 (4) | 158 (3) | |||
| Cisplatin-etoposide | ||||||
| Concurrent | 232 (3) | 37 (5) | 180 (3) | |||
| Sequential | <11 (<1) | <11 (<2) | <11 (<1) | |||
| Other | ||||||
| Concurrent | >717 (>10) | >72 (>10) | >554 (>10) | |||
| Sequential | 2232 (32) | 240 (34) | 1771 (33) | |||
| RT facility | ||||||
| Freestanding center | 1674 (24) | 115 (16) | 1346 (25) | 213 (26) | <.0001 | <.0001 |
| Hospital-based center | 2067 (30) | 308 (43) | 1514 (28) | 245 (30) | ||
| Hospital-based NCI center | 3153 (46) | 293 (41) | 2496 (47) | 364 (44) | ||
| Course length | ||||||
| 3 wk | 407 (6) | 14 (2) | 274 (5) | 119 (14) | <.0001 | <.0001 |
| 4 wk | 358 (5) | 18 (3) | 277 (5) | 63 (8) | ||
| 5 wk | 416 (6) | 32 (4) | 310 (6) | 74 (9) | ||
| 6 wk | 809 (12) | 89 (12) | 609 (11) | 111 (14) | ||
| 7 wk | 1843 (27) | 188 (26) | 1481 (28) | 174 (21) | ||
| 8 wk | 2229 (32) | 255 (36) | 1773 (33) | 201 (24) | ||
| 9 wk | 832 (12) | 120 (17) | 632 (12) | 80 (10) | ||
Abbreviations: 2D-RT = 2-dimensional radiation therapy; 3D-CRT = 3-dimensional conformal radiation therapy; IMRT = intensity modulated radiation therapy; NCI = National Cancer Institute; NSCLC = non-small cell lung cancer; PET = positron emission tomography.
Exact figures not specified in some cells to protect patient identity as indicated by SEER-Medicare data use agreement.
Characteristics associated with choice of radiation modality
Compared with 3D-CRT, patients receiving IMRT were less likely to be white and were more likely to be diagnosed in the West and in urban settings (Table 1). The IMRT cohort had higher comorbidity indices and cardiac risk factors, more home oxygen use, higher nodal stage, more PET scans and brain imaging, fewer surgical procedures, higher likelihood of RT at hospital-based centers, and longer RT course lengths.
Radiation modality and outcome
Unadjusted OS was improved with IMRT compared with 3D-CRT, with a hazard ratio (HR) of 0.90 (95% confidence interval [CI] 0.82–0.98, P = .02). CSS was also improved with IMRT, with a HR of 0.89 (95% CI 0.81–0.98, P = .02). The outcomes were inferior for 2D-RT (Fig. 1B and C).
Univariate and multivariate analyses of factors associated with OS and CSS are shown in Table 2. After adjustment, outcomes were similar between 3D-CRT and IMRT (OS HR 0.94, P = .23; CSS HR 0.94, P = .28). 2D-RT continued to be associated with inferior OS, although there was only a trend toward inferior CSS compared with 3D-CRT (HR 1.09, P = .09).
Table 2.
Univariate and multivariate adjusted models of overall and cancer-specific survival
| Overall survival | Cancer-specific survival | |||||||
|---|---|---|---|---|---|---|---|---|
|
|
|
|||||||
| Univariate | P | Multivariate* | P | Univariate | P | Multivariate | P | |
|
| ||||||||
| Likelihood ratio | 92,220 | 90,508 | 80,701 | 79,168 | ||||
| Treatment | ||||||||
| 3D-CRT | Ref | Ref | Ref | Ref | ||||
| IMRT | 0.90 (0.82–0.98) | .02 | 0.94 (0.85–1.04) | .23 | 0.89 (0.81–0.98) | .02 | 0.94 (0.85–1.05) | .28 |
| 2D-RT | 1.30 (1.19–1.42) | <.0001 | 1.13 (1.03–1.24) | .008 | 1.28 (1.17–1.41) | <.0001 | 1.09 (0.99–1.20) | .09 |
| Age | 1.03 (1.02–1.03) | <.0001 | 1.01 (1.01–1.02) | <.0001 | 1.03 (1.02–1.03) | <.0001 | 1.01 (1.01–1.02) | .0001 |
| Sex | ||||||||
| Male | Ref | Ref | Ref | Ref | ||||
| Female | 0.84 (0.80–0.89) | <.0001 | 0.85 (0.80–0.90) | <.0001 | 0.87 (0.82–0.92) | <.0001 | 0.87 (0.82–0.93) | <.0001 |
| Race | ||||||||
| White | Ref | Ref | ||||||
| Black | 1.17 (1.07–1.29) | .0006 | 1.17 (1.06–1.29) | .002 | ||||
| Hispanic | 1.04 (0.88–1.22) | .63 | 1.06 (0.90–1.26) | .48 | ||||
| Other | 0.98 (0.87–1.12) | .82 | 0.99 (0.86–1.14) | .92 | ||||
| Marital status | ||||||||
| Unmarried | Ref | Ref | ||||||
| Married | 0.92 (0.87–0.97) | .002 | 0.92 (0.87–0.98) | .006 | ||||
| Unknown | 1.04 (0.86–1.25) | .70 | 1.08 (0.88–1.32) | .46 | ||||
| Geographic area | ||||||||
| West | Ref | Ref | ||||||
| Midwest | 1.01 (0.94–1.10) | .71 | 0.99 (0.91–1.08) | .84 | 1.02 (0.93–1.11) | .71 | 0.99 (0.90–1.09) | .83 |
| South | 1.11 (1.04–1.19) | .003 | 1.09 (1.01–1.18) | .03 | 1.11 (1.03–1.19) | .005 | 1.09 (1.00–1.18) | .04 |
| Northeast | 1.05 (0.98–1.12) | .17 | 1.02 (0.95–1.10) | .52 | 1.04 (0.97–1.12) | .27 | 1.02 (0.94–1.10) | .65 |
| Urban setting | ||||||||
| Urban | Ref | Ref | Ref | |||||
| Rural | 1.04 (0.96–1.13) | .30 | 0.92 (0.83–1.01) | .09 | 1.06 (0.97–1.15) | .21 | ||
| Year of diagnosis | ||||||||
| 2002 | Ref | Ref | Ref | Ref | ||||
| 2003 | 0.92 (0.83–1.01) | .09 | 0.97 (0.87–1.08) | .58 | 0.91 (0.82–1.02) | .10 | 0.97 (0.86–1.09) | .61 |
| 2004 | 0.90 (0.82–1.00) | .05 | 0.95 (0.85–1.06) | .34 | 0.89 (0.80–0.99) | .04 | 0.94 (0.83–1.06) | .30 |
| 2005 | 0.90 (0.81–1.00) | .04 | 0.94 (0.84–1.05) | .26 | 0.87 (0.78–0.98) | .02 | 0.92 (0.82–1.04) | .18 |
| 2006 | 0.86 (0.78–0.95) | .004 | 0.86 (0.76–0.96) | .009 | 0.85 (0.76–0.94) | .003 | 0.85 (0.75–0.96) | .009 |
| 2007 | 0.83 (0.75–0.92) | .0002 | 0.95 (0.83–1.10) | .49 | 0.79 (0.71–0.88) | <.0001 | 0.91 (0.79–1.06) | .24 |
| 2008 | 0.76 (0.69–0.85) | <.0001 | 0.87 (0.75–1.01) | .06 | 0.74 (0.67–0.83) | <.0001 | 0.86 (0.73–1.00) | .05 |
| 2009 | 0.83 (0.74–0.92) | .0008 | 0.95 (0.81–1.10) | .50 | 0.82 (0.72–0.92) | .0009 | 0.94 (0.80–1.11) | .47 |
| Educational attainment of census tract or zip code | ||||||||
| Quartile 1 | Ref | Ref | Ref | Ref | ||||
| Quartile 2 | 0.91 (0.84–0.98) | .01 | 0.98 (0.91–1.06) | .65 | 0.90 (0.83–0.97) | .009 | 0.97 (0.89–1.05) | .43 |
| Quartile 3 | 0.95 (0.88–1.02) | .14 | 1.02 (0.94–1.11) | .61 | 0.96 (0.89–1.04) | .34 | 1.04 (0.96–1.13) | .35 |
| Quartile 4 | 0.84 (0.78–0.90) | <.0001 | 0.93 (0.85–1.01) | .08 | 0.85 (0.79–0.92) | <.0001 | 0.95 (0.87–1.04) | .23 |
| Median income of census tract or zip code | ||||||||
| Quartile 1 | Ref | Ref | ||||||
| Quartile 2 | 0.96 (0.89–1.03) | .29 | 0.96 (0.88–1.03) | .26 | ||||
| Quartile 3 | 0.94 (0.87–1.01) | .10 | 0.92 (0.85–1.00) | .04 | ||||
| Quartile 4 | 0.84 (0.78–0.90) | <.0001 | 0.84 (0.77–0.91) | <.0001 | ||||
| Comorbidity | ||||||||
| 0 | Ref | Ref | Ref | Ref | ||||
| 1 | 1.29 (1.20–1.38) | <.0001 | 1.19 (1.10–1.28) | <.0001 | 1.22 (1.13–1.31) | <.0001 | 1.15 (1.06–1.25) | .001 |
| 2 | 1.17 (1.06–1.30) | .002 | 1.13 (1.01–1.26) | .04 | 1.06 (0.95–1.19) | .30 | 1.06 (0.94–1.20) | .32 |
| ≥3 | 1.62 (1.40–1.88) | <.0001 | 1.30 (1.11–1.52) | .001 | 1.44 (1.22–1.69) | <.0001 | 1.20 (1.00–1.43) | .05 |
| COPD status | ||||||||
| No | Ref | Ref | Ref | Ref | ||||
| Yes | 1.20 (1.14–1.26) | <.0001 | 1.10 (1.04–1.17) | .002 | 1.14 (1.08–1.21) | <.0001 | 1.09 (1.02–1.17) | .009 |
| Oxygen-dependent status | ||||||||
| No | Ref | Ref | Ref | Ref | ||||
| Yes | 1.27 (1.16–1.39) | <.0001 | 0.97 (0.87–1.07) | .52 | 1.12 (1.01–1.24) | .03 | 0.87 (0.77–0.98) | .03 |
| Performance score | ||||||||
| 0 | Ref | Ref | Ref | Ref | ||||
| 1 | 1.15 (1.04–1.28) | .008 | 0.91 (0.81–1.03) | .13 | 1.06 (0.95–1.19) | .28 | 0.88 (0.77–0.99) | .04 |
| ≥2 | 1.40 (1.22–1.60) | <.0001 | 1.11 (0.96–1.29) | .17 | 1.27 (1.10–1.48) | .002 | 1.07 (0.90–1.26) | .45 |
| Cardiovascular risk factors | ||||||||
| 0 | Ref | Ref | Ref | Ref | ||||
| 1 | 0.98 (0.91–1.05) | .48 | 0.99 (0.92–1.07) | .81 | 0.95 (0.88–1.03) | .20 | 0.98 (0.91–1.06) | .63 |
| 2 | 0.99 (0.92–1.06) | .77 | 1.05 (0.97–1.13) | .21 | 0.95 (0.88–1.02) | .15 | 1.03 (0.95–1.12) | .50 |
| ≥3 | 0.96 (0.88–1.04) | .32 | 0.98 (0.89–1.08) | .67 | 0.89 (0.81–0.97) | .01 | 0.94 (0.85–1.04) | .24 |
| Tumor site | ||||||||
| Upper lobe | Ref | Ref | Ref | Ref | ||||
| Middle lobe | 1.01 (0.88–1.15) | .91 | 1.06 (0.92–1.21) | .43 | 1.01 (0.88–1.17) | .85 | 1.07 (0.92–1.24) | .37 |
| Lower lobe | 1.15 (1.08–1.22) | <.0001 | 1.08 (1.01–1.15) | .02 | 1.15 (1.07–1.22) | <.0001 | 1.07 (1.00–1.15) | .04 |
| Other | 1.22 (1.13–1.32) | <.0001 | 1.09 (0.99–1.19) | .08 | 1.22 (1.12–1.33) | <.0001 | 1.07 (0.97–1.18) | .16 |
| Tumor laterality | ||||||||
| Left | Ref | Ref | Ref | Ref | ||||
| Right | 1.03 (0.97–1.08) | .31 | 1.04 (0.99–1.11) | .13 | 1.05 (0.99–1.11) | .09 | 1.06 (1.00–1.13) | .05 |
| Other/both | 1.01 (0.76–1.33) | .96 | 0.80 (0.60–1.07) | .14 | 1.06 (0.79–1.41) | .71 | 0.83 (0.62–1.13) | .24 |
| Histology | ||||||||
| Adenocarcinoma | Ref | Ref | Ref | Ref | ||||
| Squamous | 1.31 (1.23–1.40) | <.0001 | 1.07 (1.00–1.15) | .04 | 1.28 (1.20–1.38) | <.0001 | 1.06 (0.99–1.14) | .11 |
| Other NSCLC | 1.27 (1.19–1.36) | <.0001 | 1.03 (0.96–1.11) | .40 | 1.27 (1.18–1.36) | <.0001 | 1.02 (0.94–1.11) | .63 |
| Grade | ||||||||
| Low | Ref | Ref | Ref | Ref | ||||
| Intermediate | 1.14 (0.96–1.34) | .13 | 1.25 (1.04–1.50) | .02 | 1.12 (0.93–1.34) | .23 | 1.22 (1.01–1.48) | .04 |
| High | 1.22 (1.04–1.44) | .01 | 1.31 (1.10–1.56) | .003 | 1.23 (1.04–1.47) | .02 | 1.32 (1.09–1.59) | .004 |
| Undifferentiated | 1.21 (0.96–1.52) | .11 | 1.20 (0.94–1.55) | .14 | 1.23 (0.96–1.56) | .10 | 1.22 (0.93–1.59) | .15 |
| Unknown | 1.32 (1.13–1.55) | .0006 | 1.29 (1.08–1.53) | .005 | 1.33 (1.12–1.58) | .001 | 1.29 (1.07–1.55) | .007 |
| Tumor size | ||||||||
| ≤2 cm | Ref | Ref | Ref | Ref | ||||
| 2.1–3 cm | 1.18 (1.05–1.32) | .006 | 1.20 (1.05–1.37) | .006 | 1.15 (1.01–1.31) | .03 | 1.18 (1.02–1.35) | .02 |
| 3.1–5 cm | 1.35 (1.21–1.49) | <.0001 | 1.35 (1.20–1.52) | <.0001 | 1.35 (1.20–1.51) | <.0001 | 1.35 (1.19–1.53) | <.0001 |
| 5.1–7 cm | 1.55 (1.39–1.73) | <.0001 | 1.54 (1.36–1.75) | <.0001 | 1.59 (1.41–1.80) | <.0001 | 1.56 (1.37–1.79) | <.0001 |
| >7 cm | 1.81 (1.61–2.04) | <.0001 | 1.77 (1.55–2.03) | <.0001 | 1.85 (1.63–2.10) | <.0001 | 1.78 (1.55–2.05) | <.0001 |
| Unknown | 1.88 (1.68–2.10) | <.0001 | 1.65 (1.44–1.88) | <.0001 | 1.90 (1.69–2.15) | <.0001 | 1.66 (1.44–1.91) | <.0001 |
| Nodal status | ||||||||
| N0 | Ref | Ref | Ref | Ref | ||||
| N1 | 1.13 (1.00–1.28) | .05 | 1.22 (1.08–1.39) | .002 | 1.18 (1.03–1.34) | .01 | 1.28 (1.12–1.47) | .0004 |
| N2 | 1.07 (1.00–1.14) | .06 | 1.20 (1.12–1.30) | <.0001 | 1.09 (1.01–1.17) | .03 | 1.21 (1.12–1.32) | <.0001 |
| N3 | 1.10 (1.00–1.20) | .04 | 1.24 (1.12–1.37) | <.0001 | 1.13 (1.03–1.25) | .01 | 1.26 (1.13–1.40) | <.0001 |
| Unknown | 1.48 (1.19–1.84) | .0004 | 1.00 (0.77–1.30) | .99 | 1.60 (1.29–2.00) | <.0001 | 1.07 (0.83–1.39) | .59 |
| PET scan | ||||||||
| Not performed | Ref | Ref | Ref | Ref | ||||
| Performed | 0.82 (0.77–0.87) | <.0001 | 0.86 (0.77–0.96) | .006 | 0.81 (0.76–0.86) | <.0001 | 0.86 (0.77–0.97) | .01 |
| Brain imaging | ||||||||
| Not performed | Ref | Ref | Ref | Ref | ||||
| Performed | 1.06 (1.00–1.11) | .04 | 1.15 (1.09–1.22) | <.0001 | 1.07 (1.01–1.13) | .01 | 1.16 (1.10–1.24) | <.0001 |
| Invasive mediastinal staging | ||||||||
| Not examined | Ref | Ref | Ref | Ref | ||||
| Examined | 0.69 (0.65–0.74) | <.0001 | 0.85 (0.79–0.92) | <.0001 | 0.70 (0.65–0.75) | <.0001 | 0.86 (0.79–0.93) | .0001 |
| Surgery | ||||||||
| None | Ref | Ref | Ref | Ref | ||||
| Sublobectomy | 0.80 (0.68–0.94) | .006 | 0.88 (0.75–1.04) | .14 | 0.83 (0.71–0.98) | .03 | 0.93 (0.78–1.10) | .41 |
| Lobectomy/pneumonectomy | 0.48 (0.44–0.52) | <.0001 | 0.49 (0.44–0.54) | <.0001 | 0.46 (0.42–0.50) | <.0001 | 0.45 (0.40–0.51) | <.0001 |
| Chemotherapy | ||||||||
| None | Ref | Ref | Ref | Ref | ||||
| Carboplatin-paclitaxel | ||||||||
| Concurrent | 0.68 (0.63–0.73) | <.0001 | 0.82 (0.75–0.89) | <.0001 | 0.67 (0.62–0.73) | <.0001 | 0.80 (0.73–0.88) | <.0001 |
| Sequential | 0.53 (0.46–0.62) | <.0001 | 0.66 (0.56–0.78) | <.0001 | 0.54 (0.45–0.63) | <.0001 | 0.65 (0.55–0.77) | <.0001 |
| Cisplatin-etoposide | ||||||||
| Concurrent | 0.51 (0.43–0.60) | <.0001 | 0.69 (0.57–0.83) | <.0001 | 0.51 (0.42–0.61) | <.0001 | 0.67 (0.55–0.81) | <.0001 |
| Sequential | 0.21 (0.06–0.73) | .01 | 0.36 (0.11–1.26) | .11 | 0.25 (0.07–0.85) | .03 | 0.44 (0.13–1.45) | .18 |
| Other | ||||||||
| Concurrent | 0.81 (0.74–0.89) | <.0001 | 0.94 (0.85–1.05) | .28 | 0.82 (0.74–0.91) | .0001 | 0.94 (0.84–1.05) | .28 |
| Sequential | 0.59 (0.56–0.63) | <.0001 | 0.64 (0.60–0.69) | <.0001 | 0.61 (0.56–0.65) | <.0001 | 0.64 (0.59–0.69) | <.0001 |
| RT facility | ||||||||
| Freestanding center | Ref | Ref | Ref | Ref | ||||
| Hospital-based center | 1.08 (1.00–1.16) | .04 | 1.07 (0.99–1.15) | .1 | 1.07 (0.99–1.16) | .08 | 1.06 (0.98–1.15) | .16 |
| Hospital-based NCI center | 0.99 (0.93–1.06) | .86 | 0.97 (0.90–1.04) | .41 | 0.99 (0.92–1.06) | .81 | 0.97 (0.90–1.04) | .39 |
| Course length | ||||||||
| 3 wk | Ref | Ref | Ref | Ref | ||||
| 4 wk | 0.78 (0.65–0.94) | .01 | 0.78 (0.65–0.94) | .009 | 0.79 (0.65–0.96) | .02 | 0.79 (0.66–0.96) | .02 |
| 5 wk | 0.45 (0.37–0.54) | <.0001 | 0.58 (0.48–0.69) | <.0001 | 0.46 (0.38–0.56) | <.0001 | 0.60 (0.50–0.73) | <.0001 |
| 6 wk | 0.29 (0.25–0.34) | <.0001 | 0.42 (0.36–0.49) | <.0001 | 0.29 (0.25–0.34) | <.0001 | 0.42 (0.36–0.49) | <.0001 |
| 7 wk | 0.28 (0.24–0.32) | <.0001 | 0.33 (0.28–0.37) | <.0001 | 0.27 (0.24–0.32) | <.0001 | 0.32 (0.28–0.37) | <.0001 |
| 8 wk | 0.29 (0.25–0.33) | <.0001 | 0.32 (0.28–0.37) | <.0001 | 0.29 (0.25–0.33) | <.0001 | 0.31 (0.27–0.36) | <.0001 |
| 9 wk | 0.34 (0.29–0.39) | <.0001 | 0.37 (0.32–0.43) | <.0001 | 0.34 (0.29–0.39) | <.0001 | 0.36 (0.31–0.42) | <.0001 |
Abbreviations: 2D-RT = 2-dimensional radiation therapy; 3D-CRT = 3-dimensional conformal radiation therapy; HR = hazard ratio; IMRT = intensity modulated radiation therapy; NCI = National Cancer Institute; NSCLC = non-small cell lung cancer; PET = positron emission tomography; RT = radiation therapy.
All multivariate models adjusted for age, year of diagnosis, comorbidity index, COPD, oxygen use, performance score, cardiac risk factors, tumor histology, tumor grade, tumor size, nodal stage, PET, brain imaging, invasive mediastinal evaluation, surgery, chemotherapy, facility type, and RT course length. Overall survival model was additionally adjusted for sex, geographic area, urban setting, area educational attainment, tumor site, and tumor laterality. Cancer-specific model was additionally adjusted for sex, geographic area, area educational attainment, tumor site, and tumor laterality.
Given that the majority of patients are now treated with 3D-CRT or IMRT (Fig. 1A), we focused the remainder of our analyses on these 2 groups. We performed propensity score matching, and the following variables did not meet balance criteria: sex, race, marital status, year of diagnosis, comorbidity score, performance status, tumor site, tumor laterality, grade, nodal status, PET scan, brain imaging, surgery, and RT facility (Table 3). All variables were well-balanced. OS and CSS continued to be similar between IMRT and 3D-CRT (Fig. 2A and B, Table 4).
Table 3.
Baseline characteristics of propensity score matched cohort (IMRT vs 3D-CRT).
| Characteristic | 3D-CRT n = 1923 (%) | IMRT n = 714 (%) | SD |
|---|---|---|---|
|
| |||
| Age | |||
| Mean ± SD | 74.4 ± 5.8 | 74.5 ± 5.9 | 0 |
| Median (range) | 74 (65–94) | 74 (65–94) | |
| Sex | |||
| Female | 46 | 47 | 1.9 |
| Race | |||
| White | 83 | 82 | −0.1 |
| Black | 8 | 9 | 1.3 |
| Hispanic | 4 | 4 | −1.7 |
| Other | 4 | 4 | 0.1 |
| Marital status | |||
| Unmarried | 42 | 42 | 0.5 |
| Married | 56 | 56 | −0.9 |
| Unknown | 2 | 2 | 1.5 |
| Geographical area | |||
| West | 37 | 38 | 1 |
| Midwest | 14 | 13 | 0.9 |
| South | 29 | 28 | −1.8 |
| Northeast | 21 | 20 | 0 |
| Urban setting | |||
| Rural | 8 | 7 | −0.8 |
| Year of diagnosis | |||
| 2002 | 3 | 2 | −0.6 |
| 2003 | 3 | 3 | −0.3 |
| 2004 | 3 | 4 | 4.1 |
| 2005 | 6 | 7 | 3.9 |
| 2006 | 11 | 10 | −0.6 |
| 2007 | 20 | 17 | −3.7 |
| 2008 | 26 | 26 | −0.2 |
| 2009 | 28 | 31 | 0.3 |
| Educational attainment of census tract or zip code | |||
| Quartile 1 | 23 | 24 | 0.2 |
| Quartile 2 | 24 | 24 | −0.3 |
| Quartile 3 | 25 | 25 | 0.9 |
| Quartile 4 | 27 | 27 | −0.7 |
| Median income of census tract or zip code | |||
| Quartile 1 | 24 | 24 | −1.2 |
| Quartile 2 | 23 | 23 | 0.7 |
| Quartile 3 | 25 | 25 | 1.6 |
| Quartile 4 | 28 | 28 | −1.1 |
| Comorbidity | |||
| 0 | 74 | 73 | −2.2 |
| 1 | 14 | 15 | 2.9 |
| 2 | 7 | 7 | −0.9 |
| ≥3 | 5 | 5 | 0.9 |
| COPD status | |||
| Yes | 35 | 37 | 2.9 |
| Oxygen-dependent status | |||
| Yes | 11 | 12 | 0.6 |
| Cardiovascular risk factors | |||
| 0 | 20 | 20 | 1.5 |
| 1 | 29 | 29 | −2.5 |
| 2 | 34 | 33 | 1 |
| ≥3 | 18 | 19 | 1.4 |
| Performance status | |||
| 0 | 86 | 86 | −1 |
| 1 | 10 | 9 | −1.5 |
| ≥2 | 4 | 5 | 1.6 |
| Tumor site | |||
| Upper Lobe | 59 | 59 | −0.9 |
| Middle Lobe | 4 | 3 | −2.4 |
| Lower Lobe | 25 | 26 | 2.7 |
| Other | 12 | 11 | −0.8 |
| Tumor laterality | |||
| Left | 44 | 44 | −1.7 |
| Right | 55 | 55 | 1.9 |
| Other/both | 1 | <2 | −1.2 |
| Histology | |||
| Adenocarcinoma | 31 | 30 | −2.6 |
| Squamous | 40 | 41 | 1.3 |
| Other NSCLC | 29 | 29 | 1.2 |
| Grade | |||
| Low | 2 | 2 | 2.2 |
| Intermediate | 17 | 16 | −2.2 |
| High | 34 | 34 | 0.3 |
| Undifferentiated | 2 | 2 | 1.4 |
| Unknown | 46 | 46 | 0.3 |
| Tumor size | |||
| ≤2 cm | 9 | 9 | 1.1 |
| 2.1–3 cm | 12 | 12 | 0.9 |
| 3.1–5 cm | 30 | 31 | 0.9 |
| 5.1–7 cm | 20 | 19 | −0.1 |
| >7 cm | 14 | 13 | −2.7 |
| Unknown | 16 | 16 | −0.2 |
| Nodal status | |||
| N0 | 18 | 18 | −1 |
| N1 | 5 | 5 | 0.9 |
| N2 | 62 | 61 | −1.4 |
| N3 | 14 | 16 | 2.4 |
| Unknown | <1 | <2 | 0.7 |
| PET scan | |||
| Performed | 60 | 62 | −1.6 |
| Brain imaging | |||
| Performed | 60 | 62 | 0.1 |
| Invasive mediastinal staging | |||
| Examined | 19 | 18 | 0.2 |
| Surgery | |||
| None | 86 | 87 | 2.9 |
| Sublobectomy | 3 | 4 | 0 |
| Lobectomy/pneumonectomy | 11 | 9 | −3.4 |
| Chemotherapy | |||
| None | 27 | 25 | −1.8 |
| Carboplatin-paclitaxel | |||
| Concurrent | 20 | 20 | 1.1 |
| Sequential | 3 | 4 | 1.6 |
| Cisplatin-etoposide | |||
| Concurrent | 5 | 5 | −2.2 |
| Sequential | <1 | <2 | 3.1 |
| Other | |||
| Concurrent | 12 | 11 | 0.1 |
| Sequential | 33 | 33 | 0.8 |
| RT facility | |||
| Freestanding center | 19 | 16 | −2.7 |
| Hospital-based center | 39 | 43 | 2.5 |
| Hospital-based NCI center | 43 | 41 | −0.4 |
| Courselength | |||
| 3 wks | 2 | 2 | 0 |
| 4 wks | 3 | 3 | −2 |
| 5 wks | 5 | 4 | −1.3 |
| 6 wks | 13 | 12 | −0.5 |
| 7 wks | 28 | 26 | −1.8 |
| 8 wks | 34 | 36 | 2.6 |
| 9 wks | 15 | 17 | 0.9 |
Abbreviations: 3D-CRT = 3-dimensional conformal radiation therapy; COPD = chronic obstructive pulmonary disease; IMRT = intensity modulated radiation therapy; NCI = National Cancer Institute; NSCLC = non-small cell lung cancer; PET = positron emission tomography; RT = radiation therapy; SD = standardized difference.
Exact figures not specified in some cells to protect patient identity.
Fig. 2.
Propensity score matched analysis of overall survival and cancer-specific survival of patients treated with 3-dimensional conformal radiation therapy (3D-CRT) and intensity modulated radiation therapy (IMRT) for stage III non-small cell lung cancer. (A) Overall survival is similar on propensity score matched analysis (multivariate adjusted proportional hazards model P =. 29). (B) Cancer-specific survival is similar on propensity score matched analysis (multivariate adjusted proportional hazards model P = .32).
Table 4.
Summary of hazard ratios for IMRT versus 3D-CRT (reference)
| Unadjusted | Multivariate adjusted | Propensity matched | ||||
|---|---|---|---|---|---|---|
|
|
|
|
||||
| HR (95% CI)* | P | HR (95% CI) | P | HR (95% CI) | P | |
|
| ||||||
| OS | 0.90 (0.82–0.98) | .02 | 0.94 (0.85–1.04) | .23 | 0.96 (0.87–1.06) | .40 |
| CSS | 0.89 (0.81–0.98) | .02 | 0.94 (0.85–1.05) | .28 | 0.96 (0.86–1.07) | .45 |
| Early UGI toxicity | 0.99 (0.85–1.14) | .88 | 1.01 (0.87–1.19) | .86 | 1.01 (0.86–1.19) | .89 |
| Late UGI toxicity | 1.11 (1.04–1.19) | .003 | 0.94 (0.87–1.01) | .08 | 0.98 (0.90–1.07) | .65 |
| Early pulmonary toxicity | 1.10 (0.90–1.36) | .34 | 1.14 (0.92–1.43) | .23 | 1.19 (0.95–1.49) | .14 |
| Late pulmonary toxicity | 1.25 (0.85–1.84) | .25 | 1.22 (0.82–1.83) | .33 | 1.38 (0.89–2.13) | .15 |
| Cardiac toxicity | 0.88 (0.66–1.18) | .40 | 0.88 (0.64–1.21) | .44 | 0.99 (0.71–1.37) | .95 |
Abbreviations: 2D-RT = 2-dimensional radiation therapy; 3D-CRT = 3-dimensional conformal radiation therapy; CI = confidence interval; CSS = cancer specific survival; HR = hazard ratio; IMRT = intensity modulated radiation therapy; OS = overall survival; RT = radiation therapy; UGI = upper gastrointestinal.
Significant values in bold (P<.05).
Radiation modality and toxicity
We next compared IMRT and 3D-CRT based on toxicity events. On univariate analysis, IMRT was associated with similar risks of early UGI toxicity (HR 0.99, P = .88), early pulmonary toxicity (HR 1.10, P = .34), late pulmonary toxicity (HR 1.25, P = .25), and cardiac toxicity (HR 0.88, P = .40). IMRT was associated with inferior risk of late UGI toxicity (HR 1.11, P = .003). On multivariate analysis, IMRT continued to be associated with similar risks of early UGI, pulmonary, and cardiac toxicities, but the association with inferior late UGI toxicity was no longer observed (HR 0.94, P = .08) (Table E2, available online at www.redjournal.org). On propensity score matching, risks of toxicities remained similar (Table 4).
Thus, IMRT is associated with similar toxicities while maintaining good cancer outcomes. All survival and toxicity models met the proportional hazards assumption and demonstrated goodness of fit.
Discussion
Although IMRT has gained acceptance in the treatment of several cancers, its use for stage III NSCLC remains somewhat controversial because of concerns about potentially inferior cancer outcomes related to interplay between MLC and tumor motion, and potentially increased toxicity caused by larger volumes of normal tissue being exposed to low-dose radiation. Because no randomized trials exist, we performed a population-based comparative effectiveness analysis of the 2 modalities with respect to both outcome and toxicity using the SEER-Medicare database. We found that use of IMRT increased significantly between 2002 and 2009, coincident with a sharp decline in the use of 2D-RT. Survival outcomes after IMRT were improved on univariate analysis compared with 3D-CRT and nonsignificantly favored IMRT after adjustment for cofounders and matching on propensity score. Compared with 3D-CRT, OS was significantly inferior and CSS was nonsignificantly inferior for 2D-RT. Our analysis confirms and extends results of a previous SEER-Medicare study showing that CT-based simulation (ie, 3D-CRT) improves outcomes over 2D simulation in patients with stage III NSCLC (20). Liao et al performed the largest single-institution retrospective study comparing IMRT with 3D-CRT, including 91 patients treated with IMRT, and found improved OS with IMRT (6). The authors suggest that the difference in OS may be a result of decreased toxicity and/or increased use of PET with IMRT (82% vs 49%) that was not accounted for in their multivariate model. We also noted improved OS and CSS on univariate analysis, but in the multivariate adjusted and propensity score matched models, outcomes were similar between the 2 groups, although the hazard ratios favored IMRT. This suggests that future analyses containing a larger number of patients may show improvement in outcomes in patients treated with IMRT.
Although IMRT allows the generation of more conformal treatment plans than does 3D-CRT, there is concern that interplay may result in underdosing of portions of the gross tumor volume (4, 5). However, these effects are likely small (21–23), and our finding that OS and CSS were not inferior for IMRT supports the idea that interplay likely does not result in clinically significant effects. Thus, the use of IMRT in patients with locally advanced NSCLC appears to be as effective as 3D-CRT.
Planning studies have found that the increased conformity of IMRT can allow for decreased doses to the esophagus and/or lung (2, 3, 24, 25), and a study by Yom et al that was expanded upon by Liao et al found lower levels of grade ≥3 pneumonitis in patients treated with IMRT (6, 7). We did not find differences in pulmonary toxicity. However, pulmonary toxicity is difficult to ascertain from the SEER-Medicare data because events are identified from diagnosis codes, which are less reliable than procedure claims. As for esophageal toxicity, rates of clinically significant acute esophagitis are also imperfectly recorded in SEER-Medicare data. In our institutional experience we have observed lower rates of acute esophagitis in patients treated with IMRT than with 3D-CRT, similar to results from a single-institution retrospective comparison of IMRT and 3D-CRT for stage I-III small cell lung cancer that found decreased use of tube feeds with IMRT but similar rates of intravenous hydration (26). Additionally, recently reported results from Radiation Therapy Oncology Group (RTOG) 0617 demonstrate that baseline patient-reported quality of life (QOL) correlated with OS and that higher QOL scores were observed after treatment with IMRT than with 3D-CRT for stage III disease (27). This suggests that a comprehensive comparison of IMRT versus 3D-CRT would ideally include QOL data, which are not available in SEER-Medicare analyses. Taken together, the results from our study and RTOG 0617 suggest that use of IMRT could lead to improved QOL without compromising patient survival.
The weaknesses of our study are similar to those of other population-based analyses and include that even after controlling for available demographic and clinical variables, there is a potential for unmeasured confounding. For example, SEER-Medicare data do not include information on important RT-related variables such as radiation dose, use of dose escalation, dose-volume histogram parameters, respiratory motion management strategies, and use of elective nodal irradiation (ENI). The data also do not contain information on many toxicity management techniques such as the use of medication for controlling esophagitis that would help to differentiate grade of toxicity. To attempt to partially account for potential differences in prescription dose, we controlled for course length. As for ENI, a previous study found that use of PET was a potential surrogate for decreased use of ENI (28), so our inclusion of PET in the multivariate and propensity score matched models may have partially corrected for this unmeasured variable. Given these limitations, toxicities and survival outcomes should continue to be studied.
In summary, we found that use of IMRT in locally advanced NSCLC did not compromise cancer outcomes and led to similar toxicities compared with 3D-CRT.
Supplementary Material
Supplementary material for this article can be found at www.redjournal.org.
Acknowledgments–
This study used the linked SEER-Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the Applied Research Program, National Cancer Institute; the Office of Research, Development and Information, CMS; Information Management Services (IMS), Inc; and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database.
Research support from Varian Medical Systems and a Stanford Society of Physician Scholars grant helped support the purchase of SEER-Medicare data. J.P.H. was supported by the Stanford NIH/NCRR CTSA grant number TL1RR025742 and the Stanford Medical Scholars Fellowship.
Footnotes
Conflict of interest: B.W.L. has received speaking honoraria from Varian Medical Systems. J.D.M., Q.L., B.W.L., and M.D. have received research support from Varian Medical Systems. The authors report no other conflict of interest.
Presented in part as a poster and abstract at the Translational Science 2013 meeting in Washington, DC, on April 17–19, 2013.
References
- 1.Ettinger DS, Akerley W, Borghaei H, et al. Non-small cell lung cancer. J Natl Compr Canc Netw 2012;10:1236–1271. [DOI] [PubMed] [Google Scholar]
- 2.Liu HH, Wang X, Dong L, et al. Feasibility of sparing lung and other thoracic structures with intensity-modulated radiotherapy for non-small-cell lung cancer. Int J Radiat Oncol Biol Phys 2004;58:1268–1279. [DOI] [PubMed] [Google Scholar]
- 3.Murshed H, Liu HH, Liao Z, et al. Dose and volume reduction for normal lung using intensity-modulated radiotherapy for advanced-stage non-small-cell lung cancer. Int J Radiat Oncol Biol Phys 2004;58:1258–1267. [DOI] [PubMed] [Google Scholar]
- 4.Schwarz M, Van der Geer J, Van Herk M, et al. Impact of geometrical uncertainties on 3D CRT and IMRT dose distributions for lung cancer treatment. Int J Radiat Oncol Biol Phys 2006;65:1260–1269. [DOI] [PubMed] [Google Scholar]
- 5.Court LE, Wagar M, Ionascu D, et al. Management of the interplay effect when using dynamic MLC sequences to treat moving targets. Med Phys 2008;35:1926–1931. [DOI] [PubMed] [Google Scholar]
- 6.Liao ZX, Komaki RR, Thames HD Jr., et al. Influence of technologic advances on outcomes in patients with unresectable, locally advanced non-small-cell lung cancer receiving concomitant chemoradiotherapy. Int J Radiat Oncol Biol Phys 2010;76:775–781. [DOI] [PubMed] [Google Scholar]
- 7.Yom SS, Liao Z, Liu HH, et al. Initial evaluation of treatment-related pneumonitis in advanced-stage non-small-cell lung cancer patients treated with concurrent chemotherapy and intensity-modulated radiotherapy. Int J Radiat Oncol Biol Phys 2007;68:94–102. [DOI] [PubMed] [Google Scholar]
- 8.Sura S, Gupta V, Yorke E, et al. Intensity-modulated radiation therapy (IMRT) for inoperable non-small cell lung cancer: The Memorial Sloan-Kettering Cancer Center (MSKCC) experience. Radiother Oncol 2008;87:17–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Jiang ZQ, Yang K, Komaki R, et al. Long-term clinical outcome of intensity-modulated radiotherapy for inoperable non-small cell lung cancer: The MD Anderson experience. Int J Radiat Oncol Biol Phys 2012;83:332–339. [DOI] [PubMed] [Google Scholar]
- 10.Warren JL, Klabunde CN, Schrag D, et al. Overview of the SEER-Medicare data: Content, research applications, and generalizability to the United States elderly population. Med Care 2002;40. IV-3–18. [DOI] [PubMed] [Google Scholar]
- 11.Beahrs OH, Henson DE, Hutter RVP, et al. Manual for Staging of Cancer. Philadelphia: Lippincott; 1988. [Google Scholar]
- 12.Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613–619. [DOI] [PubMed] [Google Scholar]
- 13.Klabunde CN, Potosky AL, Legler JM, et al. Development of a comorbidity index using physician claims data. J Clin Epidemiol 2000;53:1258–1267. [DOI] [PubMed] [Google Scholar]
- 14.Davidoff AJ, Tang M, Seal B, et al. Chemotherapy and survival benefit in elderly patients with advanced non-small-cell lung cancer. J Clin Oncol 2010;28:2191–2197. [DOI] [PubMed] [Google Scholar]
- 15.Zhu J, Sharma DB, Gray SW, et al. Carboplatin and paclitaxel with vs without bevacizumab in older patients with advanced non-small cell lung cancer. JAMA 2012;307:1593–1601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Zhu J, Sharma DB, Chen AB, et al. Comparative effectiveness of three platinum-doublet chemotherapy regimens in elderly patients with advanced non-small cell lung cancer. Cancer 2013;119:2048–2060. [DOI] [PubMed] [Google Scholar]
- 17.Werner-Wasik M, Paulus R, Curran WJ Jr., et al. Acute esophagitis and late lung toxicity in concurrent chemoradiotherapy trials in patients with locally advanced non-small-cell lung cancer: Analysis of the Radiation Therapy Oncology Group (RTOG) database. Clin Lung Cancer 2011;12:245–251. [DOI] [PubMed] [Google Scholar]
- 18.Austin PC. Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharm Stat 2011;10:150–161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Austin PC. Assessing balance in measured baseline covariates when using many-to-one matching on the propensity-score. Pharmacoepidemiol Drug Saf 2008;17:1218–1225. [DOI] [PubMed] [Google Scholar]
- 20.Chen AB, Neville BA, Sher DJ, et al. Survival outcomes after radiation therapy for stage III non-small-cell lung cancer after adoption of computed tomography-based simulation. J Clin Oncol 2011;29:2305–2311. [DOI] [PubMed] [Google Scholar]
- 21.Bortfeld T, Jokivarsi K, Goitein M, et al. Effects of intra-fraction motion on IMRT dose delivery: Statistical analysis and simulation. Phys Med Biol 2002;47:2203–2220. [DOI] [PubMed] [Google Scholar]
- 22.Chui CS, Yorke E, Hong L. The effects of intra-fraction organ motion on the delivery of intensity-modulated field with a multileaf collimator. Med Phys 2003;30:1736–1746. [DOI] [PubMed] [Google Scholar]
- 23.Jiang SB, Pope C, Al Jarrah KM, et al. An experimental investigation on intra-fractional organ motion effects in lung IMRT treatments. Phys Med Biol 2003;48:1773–1784. [DOI] [PubMed] [Google Scholar]
- 24.Grills IS, Yan D, Martinez AA, et al. Potential for reduced toxicity and dose escalation in the treatment of inoperable non-small-cell lung cancer: A comparison of intensity-modulated radiation therapy (IMRT), 3D conformal radiation, and elective nodal irradiation. Int J Radiat Oncol Biol Phys 2003;57:875–890. [DOI] [PubMed] [Google Scholar]
- 25.Chapet O, Fraass BA, Ten Haken RK. Multiple fields may offer better esophagus sparing without increased probability of lung toxicity in optimized IMRT of lung tumors. Int J Radiat Oncol Biol Phys 2006;65:255–265. [DOI] [PubMed] [Google Scholar]
- 26.Shirvani SM, Juloori A, Allen PK, et al. Comparison of 2 common radiation therapy techniques for definitive treatment of small cell lung cancer. Int J Radiat Oncol Biol Phys 2013;87:139–147. [DOI] [PubMed] [Google Scholar]
- 27.Movsas B, Hu C, Sloan J, et al. Quality of life (QOL) analysis of the randomized radiation (RT) dose escalation NSCLC trial (RTOG 0617): The rest of the story. Int J Radiat Oncol Biol Phys 2013;87:S1–S2. [Google Scholar]
- 28.Fernandes AT, Shen J, Finlay J, et al. Elective nodal irradiation (ENI) vs. involved field radiotherapy (IFRT) for locally advanced non-small cell lung cancer (NSCLC): A comparative analysis of toxicities and clinical outcomes. Radiother Oncol 2010;95:178–184. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.