Cost-effectiveness of stereotactic body radiation therapy versus surgical resection for stage I non–small cell lung cancer


  • Anand Shah MD, MPH,

    1. Department of Radiation Oncology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania
    2. Department of Radiation Oncology, Columbia University Medical Center, New York, New York
    Search for more papers by this author
  • Stephen M. Hahn MD,

    1. Department of Radiation Oncology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Robert L. Stetson MHA,

    1. Office of Strategic Decision Support, University of Pennsylvania Health System, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Joseph S. Friedberg MD,

    1. Department of Surgery, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Taine T. V. Pechet MD,

    1. Department of Surgery, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania
    Search for more papers by this author
  • David J. Sher MD, MPH

    Corresponding author
    1. Department of Radiation Oncology, Rush University Medical Center, Chicago, Illinois
    • Corresponding author: Anand Shah, MD, MPH, Department of Radiation Oncology, Columbia University Medical Center, 622 West 168th Street, New York, NY 10032; Fax: (212) 305-5935;

    Search for more papers by this author

  • An abstract of this work has been accepted for oral presentation at the annual meeting of the American Society for Radiation Oncology in September 2013.

  • We thank Brian Wildt (Radiation Oncology), Quiana Mack (Surgery), and Mary Ferroni (Anesthesia) at the University of Pennsylvania for their assistance with Medicare coding and payment data.



The traditional treatment for clearly operable (CO) patients with stage I non–small cell lung cancer (NSCLC) is lobectomy, with wedge resection (WR) and stereotactic body radiation therapy (SBRT) serving as alternatives in marginally operable (MO) patients. Given an aging population with an increasing prevalence of screening, it is likely that progressively more people will be diagnosed with stage I NSCLC, and thus it is critical to compare the cost-effectiveness of these treatments.


A Markov model was created to compare the cost-effectiveness of SBRT with WR and lobectomy for MO and CO patients, respectively. Disease, treatment, and toxicity data were extracted from the literature and varied in sensitivity analyses. A payer (Medicare) perspective was used.


In the base case, SBRT (MO cohort), SBRT (CO cohort), WR, and lobectomy were associated with mean cost and quality-adjusted life expectancies of $42,094/8.03, $40,107/8.21, $51,487/7.93, and $49,093/8.89, respectively. In MO patients, SBRT was the dominant and thus cost-effective strategy. This result was confirmed in most deterministic sensitivity analyses as well as probabilistic sensitivity analysis, in which SBRT was most likely cost-effective up to a willingness-to-pay of more than $500,000/quality-adjusted life year. For CO patients, lobectomy was the cost-effective treatment option in the base case (incremental cost-effectiveness ratio of $13,216/quality-adjusted life year) and in nearly every sensitivity analysis.


SBRT was nearly always the most cost-effective treatment strategy for MO patients with stage I NSCLC. In contrast, for patients with CO disease, lobectomy was the most cost-effective option. Cancer 2013;119:3123–3132. © 2013 American Cancer Society.


Lung cancer is the leading cause of cancer mortality in the United States and, given the positive results of the National Lung Screening Trial (NLST),[1] it is likely that more people will be diagnosed with early-stage lung cancer in future years. The traditional treatment for patients with stage I non–small cell lung cancer (NSCLC) is lobectomy. Sublobar resection (eg, wedge resection) or ablation are often performed as an alternative to lobectomy in patients with smaller, peripheral tumors and compromised lung function. Sublobar resection has traditionally been considered inferior to anatomic lobectomy, although prospective multicenter trials are underway.

For patients who are medically inoperable, stereotactic body radiation therapy (SBRT) has effectively emerged as the standard-of-care treatment. SBRT combines precise patient localization with highly conformal treatment delivery, allowing high-dose treatment in 1 to 2 weeks (usually 10 to 18 Gy per treatment for 3 to 5 sessions). Compared with conventional radiation therapy, prospective experience with SBRT has demonstrated improved local control and survival in the setting of low treatment-related toxicity.[2, 3] The results to date have been promising enough to generate 3 active randomized trials that compare SBRT with surgery.

With the expected increase in the prevalence of stage I NSCLC and competing non-invasive modalities such as SBRT, determining the cost-effectiveness of treatments for this disease is an important question for payers, patients, providers, and other stakeholders. Both lobectomy and wedge resection are more costly than SBRT, and it is unclear whether this additional surgical risk and cost yield any value. In this study, we compared the cost-effectiveness of SBRT with wedge resection and lobectomy for marginally operable (MO) and clearly operable (CO) patients, respectively.


We designed a Markov model to simulate the clinical trajectory of a 65-year-old patient with medically operable stage I NSCLC (Fig. 1). In a Markov model, hypothetical cohorts of patients progress (ie, transition) from 1 discrete, mutually exclusive health state to another. These transitions occur in fixed time increments and at defined probabilities (ie, transition probabilities). The model was created and analyzed with TreeAge Pro 2012 (TreeAge Software, Williamstown, Mass).

Figure 1.

Schema shows the Markov model comparing SBRT with surgery for stage I non–small cell lung cancer. Abbreviations: Chemo, adjuvant chemotherapy; DM, distant metastasis; LR, local recurrence; NED, no evidence of disease; RR, regional recurrence; SBRT, stereotactic body radiotherapy; Wedge, wedge resection.


Three treatment strategies were compared. For patients who are MO, SBRT and wedge resection were compared. For patients who are CO, the cost-effectiveness of SBRT and lobectomy were compared. Although the efficacy of SBRT was assumed to be the same in both comparisons, the risk of toxicity was substantially greater in the MO cohort, because they are presumably more susceptible to treatment-related morbidity. Both open (ie, thoracotomy) and video-assisted thoracic surgery (VATS) approaches were considered for patients undergoing lobectomy and wedge resection.

Decision Model

We assumed all patients had a negative mediastinoscopy and no radiologic evidence of nodal metastases prior to treatment. Patients undergoing SBRT were exposed to a risk of pneumonitis (MO, grade 2 and 3; CO, grade 2 only) from months 0 to 6, lasting 6 months; there was a risk of chest wall pain from months 6 to 12, lasting 6 months. Patients were at risk for local (LR) and regional (RR) recurrence, and distant metastasis (DM) from years 0 to 5. There was a varying possibility of surgical salvage for an LR in the CO but not MO cohort; RR and DM were not curable. The long-term complications of SBRT in the CO cohort were based on the results of a well-respected tertiary care center in the Netherlands.[4]

Surgical patients were exposed to an immediate risk of death attributable to surgery (ie, 30-day postoperative mortality). They were also at risk for atrial fibrillation for 3 months, which was managed medically. Patients were at risk for dyspnea and/or pain for the first year after surgery, and then it resolved, consistent with data suggesting resolution of an overall quality-of-life decrement after surgery by 1 year[5]; this assumption was tested in sensitivity analyses. Patients following wedge resection were at risk for LR, RR, and DM for 5 years; identical to the MO cohort treated with SBRT, no recurrence was salvageable after wedge resection. Recurrence patterns after lobectomy are typically not stratified into LR versus RR, and thus lobectomy patients experienced a risk of locoregional recurrence (LRR) and DM for 5 years; neither were salvageable.

Given their presumed adequate performance status, patients undergoing lobectomy who were found to have occult lymphadenopathy were treated with adjuvant chemotherapy, per randomized data. There was a small risk of death with chemotherapy during its administration.

Disease and Treatment Assumptions

Table 1 depicts the probabilities, costs, and utilities used in the study. The probabilities were derived from recognized clinical studies. The LR rate for SBRT was obtained from a 3-year study of potentially operable patients in the Netherlands.[4] Despite retrospective evidence suggesting the LR rate for wedge resection is higher than after SBRT,[6] we assumed the same risk per Crabtree et al[7]; the LR risk after wedge resection was made equal to that of SBRT (7%) in the base model. The LRR rates following lobectomy were based on a series of patients with pathologically confirmed stage IA disease (no nodal involvement)[8] and the International Adjuvant Lung Cancer Trial (occult nodal disease).[9]

Table 1. Probabilities, Utility Values, and Costsa
EventBaseline ValueRange StudiedReference
  1. Costs are expressed in 2012 US dollars.

  2. Abbreviations: DM, distant metastasis; LR, local recurrence; LRR, locoregional recurrence; N0-N1-N2, nodal status; NED, no evidence of disease; SBRT, stereotactic body radiotherapy.

NED to LR   
SBRT7% (5 y)5%-20%[4]
Wedge resection7% (5 y)2%-35%[6]
NED to LRR (lobectomy only)   
N0 disease6.7% (4.25 y)3%-20%[8]
N1-N2 disease28% (4.25 y)10%-50%[9]
LR to death70% (1 y)50%-100%[18]
RR to death70% (1 y)50%-100%[18]
DM to death70% (1 y)50%-100%[18]
Death from surgery1%0.05%-5%[19]
Death from chemotherapy0.08%0.05%-3%[20]
Chest wall pain (SBRT only)20%0%-30%[18]
Grade 2 pneumonitis rate (SBRT only)24%0%-30%[3]
Grade 3 pneumonitis rate (SBRT only)12%0%-30%[3]
Atrial fibrillation (surgery only)14.4%5%-25%[19]
Postoperative pain50%0%-50%
Postoperative dyspnea (lobectomy)50%0%-50%
Postoperative dyspnea (wedge resection)13.1%0%-50%[21]
Pain control$26.54/mo[22]
Grade 2 pneumonitis$96.50/mo[22]
Grade 3 pneumonitis (one-time oxygen equipment cost, monthly cost)$1,466.27, $291.30/mo[22, 23]
Atrial fibrillation (90-day warfarin + anticoagulation monitoring)$112.27[24]
Palliative care (for any recurrence)$82,467.74$10,000-$100,000[25]
Utility values
Atrial fibrillation0.7080.461-0.825[24]
Chest wall pain0.5570.3-0.825[10]
LR, RR, LRR, or DM0.4610.1-0.557[10]

For all treatment modalities, the underlying risks of RR and DM were identical; it is the same population of patients. Thus, the risk of occult nodal disease detected in lobectomy served as the baseline risk of RR after SBRT or wedge resection. Similarly, the probability of DM was stratified by occult nodal stage (node-positive versus node-negative) that was the same in all patients. However, chemotherapy mitigated the risk of DM in lobectomy patients per International Adjuvant Lung Cancer Trial data.


Health state utilities of patients with early-stage lung cancer have not been well studied. We therefore adopted utilities from a study of health states associated with NSCLC[10] in our baseline model and tested a wide range of assumptions in sensitivity analyses. For the no-evidence-of-disease health state, we assumed a baseline utility and the presence of pain and/or dyspnea resulted in a disutility subtracted from this value for the first year; the decrement was derived from a linear regression in a prior study and tested in sensitivity analyses.[10]

Medicare Payment Data

A payer (Medicare) perspective was used in this study. Medicare 2012 payment levels for SBRT and surgery-related costs are shown in Table 2. We assumed that all treatments were performed at the University of Pennsylvania (Philadelphia, Pa), because Medicare payments vary by hospital type and location. All patients underwent outpatient SBRT, whereas surgery was performed on an inpatient basis. Outpatient hospital services, such as SBRT, are paid according to the Medicare outpatient prospective payment system (OPPS) schedule. The total cost of linear accelerator–based SBRT was $8,900.43 whereas robotic SBRT was $14,821.21; the latter was used in this study. Briefly, patients undergoing surgery are assigned a Medicare Severity–Diagnosis-Related Group (MS-DRG) based on age, diagnosis, procedures performed, comorbidities, and complications. The hospital receives a bundled payment for the patient's MS-DRG, in addition to professional fees such as those for the surgeon and anesthesiologist. Accordingly, we modeled the following current procedural terminology codes for surgeon's professional fees: 32505, 32666, 32480, and 32663 for open wedge resection, VATS wedge resection, open lobectomy, and VATS lobectomy, respectively.

Table 2. Costs Associated With SBRT and Surgerya
ActivityCPT/ICDDescriptionNo.OPPS ($)MPFS ($)IPPS ($)
  1. a

    Medicare payment levels were obtained on March 31, 2012. MPFS values are for January to December 2012 on an inpatient basis. IPPS values are for fiscal year 2012 and include patient copay.

  2. Abbreviations: CPT, current procedural terminology; CT, computed tomography; ICD, International Classification of Diseases; IPPS, inpatient prospective payment system; MPFS, Medicare physician fee schedule; MS-DRG, Medicare Severity–Diagnosis-Related Group; OPPS, outpatient prospective payment system; SBRT, stereotactic body radiotherapy; VATS, video-assisted thoracoscopic surgery.

 77334Immobilization device–complex1207.9263.22
Physician planning77263Clinical treatment planning–complex1167.80
 77470Special treatment procedure1409.91107.58
Physics plan77295Three-dimensional planning1990.77235.89
 77300Dosimetry calculation7111.7632.07
 77334Treatment device–complex7207.9263.22
TreatmentG0339Stereo body robotic treatment (first)13501.65
 G0340Stereo body robotic treatment (second-third)22615.98
Management77435Stereotactic body treatment management1645.72
SBRT Total   14821.21  
Surgical Resection
Professional Fees      
Open lobectomy32480Removal of lung, other than total pneumonectomy; single lobe (lobectomy)1690.04
VATS lobectomy32663Thoracoscopy, surgical; with lobectomy, single lobe1596.66
Open wedge resection32505Thoracotomy; with therapeutic wedge resection, initial1053.50
VATS wedge resection32666Thoracoscopy surgical; with therapeutic wedge resection, initial985.94
Anesthesia00541Anesthesia for intrathoracic procedures, excluding procedures on the heart, great vessels, trachea; utilizing one-lung ventilation (healthy low-risk patient with 90 minutes anesthesia time)1482.79
Arterial line36620Insertion of an intra-arterial monitoring line during surgery154.59
Epidural62318Catheter placement, infusion or intermittent bolus; epidural or subarachnoid; cervical or thoracic1107.73
CT thorax without contrast71250Outpatient CT thorax without contrast1448.4652.39
163 Major chest procedures with major complication or comorbidity (MCC)46210.85
164 Major chest procedures with complication or comorbidity (CC)23626.44
165 Major chest procedures without CC/MCC16206.02

MS-DRGs 163 to 165 are associated with major chest procedures. In order to estimate a case mix of patients undergoing open versus VATS surgical procedures, as well as the frequency of MS-DRG assignment, we modeled the 2011 case-mix from the 18-county Philadelphia media market less Delaware, because the latter state's data were unavailable. The 18-county MS-DRG prevalence was derived from information made available through the following: Pennsylvania Health Care Cost Containment Council, New Jersey Department of Health and Human Services, Maryland Health Services Cost Review Commission, and the New York State Planning and Research Cooperative System. The following International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) primary diagnosis codes were used: 162.2, 162.3, 162.4, 162.5, 162.8, 162.9, and 231.2.[11] These data were restricted to the following ICD-9 primary procedure codes: 32.29 (open wedge resection), 32.20 (VATS wedge resection), 32.49 (open lobectomy), 32.41 (VATS lobectomy). The average MS-DRG payment was a weighted average of the MS-DRG payments in this case-mix.

All costs were inflated to 2012 US dollars using the Consumer Price Index if necessary.[12] Costs and quality-adjusted life years (QALYs) were both discounted at 3% annually.

Sensitivity Analyses

The effects of variability in the model assumptions were evaluated using sensitivity analyses (SAs). One-way SA was performed for the parameters depicted in Table 1. One treatment strategy is considered to dominate the other approaches when it is both more effective and less costly. We report the incremental cost-effectiveness ratio (ICER) when one strategy is more effective but more costly (or less effective but less costly) compared with another. A treatment strategy with an ICER value less than the societal willingness-to-pay (WTP) is considered to be cost-effective. Societal WTP was considered to be $50,000/QALY in this study.[13]

To investigate the relationship between surgical morbidity and cost-effectiveness, we performed a series of SAs that varied the likelihood of permanent dyspnea and/or pain. After the baseline assumption of no surgical morbidity after 1 year, we tested whether the results changed if 50% and 100% of patients who developed these symptoms experienced them for life, as well as if the magnitude of disutility increased or decreased.

The model results were relatively inelastic on these deterministic SA, and thus we only performed one probabilistic sensitivity analysis (PSA) for the wedge resection and SBRT comparison. We made the highly conservative assumption that the local control rate after wedge resection was at a minimum equal to SBRT and may be up to 35% better, per the assumptions of noninferiority from the Radiation Therapy Oncology Group 1021 trial. We varied this ratio between 0.65 to 1 on a uniform distribution. Furthermore, we varied the cost of wedge resection by changing the probability of billing MS-DRG codes 164 and 165; to weight the analysis against SBRT, we assumed the probability of MS-DRG 165 (least expensive) varied on a uniform distribution between 50% and 75%. Each distribution was drawn 50,000 times, and an acceptability curve was created.


Model Validity

The 3 modeled treatment options were calibrated for consistency with established clinical studies. The model predicted 5-year LR, RR, and DM rates for patients undergoing SBRT of 7%, 11%, and 17%, respectively. The model predicted 5-year LR, RR, and DM rates for patients undergoing wedge resection of 7%, 9%, and 15%, respectively. The model predicted 5-year LRR and DM rates for patients undergoing lobectomy for node-negative disease of 6% and 12%, respectively (for node-positive disease, the respective rates were 20% and 39%). These results, although not surprising, are reassuring that the model was accurately describing the disease processes.

Base Case

SBRT-MO, SBRT-CO, wedge resection, and lobectomy were associated with a mean cost and quality-adjusted life expectancy of $42,094/8.03, $40,107/8.21, $51,487/7.93, and $49,093/8.89, respectively. In patients who are MO, SBRT was the dominant strategy and thus the most cost-effective. For patients who are CO, lobectomy was the cost-effective treatment option (ICER = $13,200/QALY).

When an open-only surgical approach was considered in the base model, wedge resection and lobectomy were associated with a mean cost of $53,570 and $49,428, respectively. Similarly, wedge resection and lobectomy were associated with a mean cost of $50,669 and $48,713, respectively, when a VATS-only analysis was performed; the ICERs were essentially unchanged. Given the relative inelasticity of the model to variability in surgical technique, the base model consisting of both open and VATS procedures was used in the subsequent analyses described below.

One-Way Sensitivity Analyses: SBRT-MO Versus Wedge Resection

Table 3 depicts the results of 1-way SA comparing SBRT-MO with wedge resection. In almost any scenario, SBRT was the dominant (and thus the most cost-effective) strategy compared with wedge resection. Similarly, SBRT remained borderline cost-effective when the cost associated with wedge resection was only $10,000 (ICER = $57,000/QALY). Wedge resection did become the cost-effective strategy when its 5-year risk of LR was 2% (ICER = $18,400/QALY) or the LR risk associated with SBRT was 20% (ICER = $5500/QALY). It is worth noting that such efficacy values are rarely if ever reported in the recent literature for stage I lung cancer, supporting the cost-effectiveness of SBRT.

Table 3. One-Way Sensitivity Analysis: SBRT Versus Wedge Resectiona
ParameterRange StudiedICER
Lower BoundUpper Bound
  1. a

    Dominance (D) refers to a strategy that is both more effective and less costly.

  2. Costs are expressed in 2012 US dollars.

  3. Abbreviations: D, dominant; LR, local recurrence; LRR, locoregional recurrence; N0, N1-N2, nodal status; NED, no evidence of disease; Q, quality-adjusted life year (QALY); SBRT, stereotactic body radiotherapy.

NED to LR  
SBRT5%-20%SBRT (D)Wedge ($5500/Q)
Wedge resection2%-35%Wedge ($18,400/Q)SBRT (D)
N1/N2 disease prevalence5%-50%SBRT (D)SBRT (D)
Death from surgery0.5%-5%SBRT (D)SBRT (D)
Chest wall pain rate (SBRT)0%-30%SBRT (D)SBRT (D)
Grade 2 pneumonitis rate (SBRT)0%-30%SBRT (D)SBRT (D)
Grade 3 pneumonitis rate (SBRT)0%-30%SBRT (D)SBRT (D)
Atrial fibrillation rate (surgery)5%-25%SBRT (D)SBRT (D)
Postoperative pain rate0%-50%SBRT (D)SBRT (D)
Postoperative dyspnea rate0%-50%SBRT (D)SBRT (D)
SBRT$8,000-$50,000SBRT (D)SBRT ($257,900/Q)
Wedge resection$10,000-$50,000SBRT ($57,000/Q)SBRT (D)
Palliative care$10,000-$100,000SBRT (D)SBRT (D)
Utility values
NED0.6-0.95SBRT (D)SBRT (D)
Chest wall pain (SBRT)0.3-0.825SBRT (D)SBRT (D)
Pneumonitis (SBRT; grade 2 and grade 3)0.461-0.825SBRT (D)SBRT (D)
Atrial fibrillation (surgery)0.461-0.825SBRT (D)SBRT (D)
Recurrence0.1-0.557SBRT (D)SBRT (D)

One-Way Sensitivity Analyses: SBRT-CO Versus Lobectomy

Under every assumption in Table 4, lobectomy was more cost-effective compared with SBRT for patients who are CO. The ICER for lobectomy was below $50,000/QALY, well below any accepted societal WTP. Lobectomy was the clearly dominant strategy when the prevalence of nodal disease (N1 or N2) was 50%, cost of SBRT was $50,000, or cost of lobectomy was $10,000. None of these scenarios are likely, however.

Table 4. One-Way Sensitivity Analysis: SBRT Versus Lobectomy (ie, Lobe)a
ParameterRange StudiedICER
Lower BoundUpper Bound
  1. a

    Dominance (D) refers to a strategy that is both more effective and less costly.

  2. Costs are expressed in 2012 US dollars.

  3. Abbreviations: D, dominant; LR, local recurrence; LRR, locoregional recurrence; N0, N1-N2, nodal status; NED, no evidence of disease; Q, quality-adjusted life year (QALY); SBRT, stereotactic body radiotherapy.

NED to LR   
SBRT5%-20%Lobe ($19,900/Q)Lobe ($4200/Q)
Lobectomy (N0)3%-20%Lobe ($7,400/Q)Lobe ($6,400/Q)
Lobectomy (N1/N2)10%-50%Lobe ($10,200/Q)Lobe ($18,400/Q)
N1/N2 disease prevalence5%-50%Lobe ($38,600/Q)Lobe (D)
Death from surgery0.5%-5%Lobe ($12,600/Q)Lobe ($23,700/Q)
Death from chemotherapy0.5%-3%Lobe ($13,400/Q)Lobe ($19,100/Q)
Surgical salvage rate10%-60%Lobe ($10,000/Q)Lobe ($19,100/Q)
Chest wall pain rate (SBRT)0%-30%Lobe ($13,400/Q)Lobe ($12,700/Q)
Grade 2 pneumonitis rate (SBRT)0%-30%Lobe ($13,200/Q)Lobe ($11,700/Q)
Grade 3 pneumonitis rate (SBRT)0%-30%Lobe ($13,200/Q)Lobe ($8,200/Q)
Atrial fibrillation rate (surgery)5%-25%Lobe ($13,000/Q)Lobe ($13,200/Q)
Postoperative pain rate0%-50%Lobe ($12,700/Q)Lobe ($13,500/Q)
Postoperative dyspnea rate0%-50%Lobe ($12,800/Q)Lobe ($13,400/Q)
SBRT$8,000-$50,000Lobe ($23,200/Q)Lobe (D)
Lobectomy$10,000-$50,000Lobe (D)Lobe ($46,000/Q)
Chemotherapy$10,000-$50,000Lobe ($9,100/Q)Lobe ($16,500/Q)
Palliative care$10,000-$100,000Lobe ($22,800/Q)Lobe ($10,900/Q)
Utility values
NED0.6-0.95Lobe ($20,900/Q)Lobe ($11,100/Q)
Chest wall pain (SBRT)0.3-0.825Lobe ($12,800/Q)Lobe ($13,400/Q)
Pneumonitis (SBRT; grade 2 and grade 3)0.461-0.825Lobe ($13,200/Q)Lobe ($13,200/Q)
Atrial fibrillation (surgery)0.461-0.825Lobe ($13,200/Q)Lobe ($13,000/Q)
Recurrence0.1-0.557Lobe ($12,300/Q)Lobe ($13,400/Q)

Two-Way Sensitivity Analysis on Surgical Morbidity

We performed a 2-way SA in which we varied both the probability that dyspnea and pain were permanent, as well as the disutility associated with them (ranging between 50% to 200% for the assumed disutility). In the MO comparison, SBRT was still the dominant strategy, even with the assumption of no permanent morbidity and a small disutility for pain and dyspnea.

In the CO comparison, lobectomy was cost-effective versus SBRT (ie, ICER below $50,000/QALY) in nearly every scenario except the most extreme: permanent pain and dyspnea, with a disutility twice that of the base case. This case resulted in an ICER of lobectomy of $90,000/QALY.

Probabilistic Sensitivity Analysis

The PSA assumed 2 conditions favorable to wedge resection: its local control rate relative to SBRT varied between 0.65 and 1, and its MS-DRG payment was the lowest possible between 50% and 75% of cases. The results of this PSA are shown in Figure 2. Even with these favorable assumptions, SBRT was most likely to be the cost-effective strategy up to a WTP well beyond $500,000/QALY.

Figure 2.

Acceptability curve is shown from probabilistic sensitivity analysis of stereotactic body radiotherapy (SBRT) versus wedge resection. The curve represents the proportion of hypothetical trials that result in an incremental cost-effectiveness ratio for SBRT less than the societal willingness-to-pay (ie, the probability of SBRT as a cost-effective strategy depicted as a function of societal willingness-to-pay, in US dollars per quality-adjusted life year [QALY]).


We have shown that for MO patients, SBRT is nearly always the dominant strategy in comparison with wedge resection, whereas in patients for whom lobectomy is a viable option, surgery is clearly cost-effective versus SBRT. These findings were robust over a wide range of assumptions including treatment efficacies, toxicities, costs, and health state utilities.

In analyzing the drivers of these results, we found that the extra cost and morbidity of wedge resection without a clear disease-control benefit rendered it a vastly inferior strategy. Even in a PSA that heavily favored wedge resection, SBRT was more likely to be cost-effective up to a societal WTP of $500,000/QALY. In contrast, surgical lobectomy, although more costly than SBRT, is the superior oncologic procedure. Wide margins are typically obtained from the resection, and the pathologic staging and lymphadenectomy further allows for optimal locoregional control and decisions on systemic therapy. The cost and upfront morbidity were easily counterbalanced by the increased cost and late morbidity of LRR after SBRT.

Both SBRT and wedge resection continue to be offered to MO patients because there is a paucity of randomized evidence strongly favoring either.[6] Moreover, there are emerging clinical data examining SBRT in CO patients.[4, 14] Our results strongly suggest cost-effective strategies for both populations. Importantly, these findings could impact a large number of patients with stage I NSCLC. Although lung cancer incidence in the United States has decreased over the past decade, there is renewed focus on disease screening. The recent results of the NLST favor low-dose computed tomography as a first-line lung cancer screening test for high-risk patients. The National Comprehensive Cancer Network has since recommended low-dose computed tomography screening for select patients. It is estimated that nearly 9 million Americans would be eligible for screening based on NLST eligibility criteria.[15] It is therefore imperative that cost-effective treatment strategies for NSCLC continue to be evaluated.

Although this is the first formal Markov-based cost-effectiveness analysis comparing surgery with SBRT, our results do echo recently published studies analyzing a similar question. Compared with 3-dimensional conformal therapy and radiofrequency ablation, SBRT has been demonstrated to be the most cost-effective nonsurgical treatment for inoperable patients.[16] Puri et al used clinical data from a small, propensity-matched cohort of patients treated with SBRT and surgery (any type of surgery) and showed that surgery was generally cost-effective, although professional fees, costs of recurrence, and QALYs were not included in the analysis.[17] Our analysis is largely concordant with these results, because SBRT dominated the significantly less effective surgical modality (ie, wedge resection) but was a markedly inferior strategy in comparison with lobectomy.

We acknowledge several potential limitations of this model. This Markov cost-effectiveness analysis is based on key assumptions, notably those related to efficacy, cost, and utility outcomes. The best available clinical outcomes data on SBRT for NSCLC, for example, are based on the results of retrospective and phase 2 data and thus the body of evidence is still maturing. We have accounted for the potential variation in outcomes across published studies by testing a wide range of assumptions through sensitivity analyses, and we biased our results against a novel technology: SBRT. Consequently, the 5-year LR, RR, and DM rates resulting from the model (which suggest slightly higher recurrence rates for SBRT compared with wedge resection) are due to conservative input assumptions and do not necessarily reflect real-world outcomes. Second, in order to minimize heterogeneity in our payer perspective, Medicare payment data from only one hospital system was used. Thus, our results are probably, but not necessarily, generalizable beyond the catchment area studied. Finally, although we adopted utilities from a study of health states associated with NSCLC, the health state utilities of patients with early-stage disease have not been studied. Future studies should evaluate patient preferences in this population.

In conclusion, compared with wedge resection, SBRT is almost always the cost-effective treatment strategy for MO patients with early-stage NSCLC. In contrast, lobectomy is typically the most cost-effective strategy for patients with CO disease. The consistency of these results provides strong evidence that in a cost-conscious health care system, SBRT and lobectomy are clearly the preferred treatment options for medically compromised and medically fit patients, respectively.


This work was supported by a health services and economics research fellowship from the University of Pennsylvania Department of Radiation Oncology (to A. Shah).


Dr. Shah is a member of the Government Relations Committee of the American Society of Clinical Oncology and the Government Relations Council (Congressional Relations Subcommittee) of the American Society for Radiation Oncology. All other authors made no disclosure.