KEY POINTS
- This national target trial emulation used England's linked cancer registry, hospital, and radiotherapy datasets to compare 12,546 patients with T1–2N0M0 non-small cell lung cancer (NSCLC) treated between 2015 and 2020. 2,299 patients received stereotactic ablative radiotherapy (SABR) and 10,247 underwent surgical resection. Eligibility was restricted to patients aged 18–79 years with Zubrod performance status 0–2, or those aged ≥80 years with performance status 0–1, to approximate a population potentially eligible for either treatment.
- Treatment strategies reflected routine national practice rather than protocolized randomized interventions. The most common SABR schedules were 55 Gy in five fractions (55.2%), 54 Gy in three fractions (21.1%), and 60 Gy in eight fractions (11.1%), while surgery predominantly involved lobectomy, with sublobar resection used in the remaining patients. Before adjustment, SABR recipients were older (median 74.9 vs 71.8 years), more frequently had performance status 2 (26.0% vs 7.4%), and had higher rates of COPD (45.8% vs 26.0%), demonstrating substantial baseline differences between the groups.
- The primary endpoint, three-year all-cause mortality, was significantly higher after SABR. Unadjusted mortality was 39.4% with SABR versus 19.8% with surgery, an absolute difference of 19.6 percentage points. Following instrumental-variable adjustment, predicted mortality was 30.3% versus 21.6%, respectively, corresponding to an adjusted absolute difference of 8.8 percentage points (95% CI 1.4–16.2; p=0.02). The instrumental variable represented historical SABR use within each multidisciplinary referral network, aiming to reduce bias from unmeasured differences influencing treatment selection.
- Cancer-specific mortality and recurrence also numerically favored surgery, but the adjusted differences were not statistically significant. Three-year lung cancer-specific mortality was 20.6% after SABR versus 14.5% after surgery, an absolute difference of 6.1 percentage points (95% CI −0.6 to 12.9; p=0.07). Corresponding three-year recurrence estimates were 34.3% versus 27.9%, with an adjusted difference of 6.5 percentage points (95% CI −1.7 to 14.6; p=0.12). Recurrence was identified from routinely collected diagnostic and treatment records rather than standardized trial surveillance.
- The survival difference was smaller and statistically inconclusive among patients aged ≥80 years. In 1,752 patients aged ≥80 with performance status 0–1, the adjusted absolute difference in three-year all-cause mortality was 4.3 percentage points (95% CI −3.7 to 12.3; p=0.62). By comparison, the difference was 10.9 points (95% CI 3.0–18.9) among patients aged 18–69 years and 8.4 points (95% CI 0.7–16.0) among those aged 70–79 years. These subgroup findings suggest a potentially different benefit-risk balance in older patients but do not establish equivalence between treatments.
- SABR was associated with fewer early hospital readmissions, although the adjusted difference was not statistically significant. The unadjusted 90-day readmission rate was 22.2% after SABR versus 30.9% after surgery, while adjusted predicted rates were 23.3% versus 30.2%, an absolute difference of −6.9 percentage points (95% CI −14.3 to 0.4). Adjusted all-cause mortality also did not differ significantly at one year (9.5% vs 6.2%; p=0.21), indicating that the mortality difference became more apparent with longer follow-up.
- Sensitivity analyses generally supported the primary finding, but residual confounding remains possible. The adjusted three-year mortality difference remained 9.2 percentage points after excluding the first COVID-19 wave and 12.7 points after excluding patients without histological confirmation, while an alternative regression-adjustment approach yielded 14.5 points. Nevertheless, 42.9% of SABR patients lacked recorded histological classification, and the instrumental-variable approach depended on assumptions about referral-network practice and unmeasured care quality that cannot be fully verified. The study also lacked patient-reported quality of life, detailed tumor anatomy, and complete functional-operability assessments.
CLINICAL TAKEAWAY
This large national observational analysis found higher adjusted three-year mortality after SABR than surgery among patients with early-stage NSCLC considered potentially suitable for either treatment, supporting surgery as the preferred curative approach for appropriately fit patients. The smaller and statistically inconclusive difference among patients aged ≥80 years reinforces the importance of individualized decisions when operative risk is substantial. However, target trial emulation and instrumental-variable adjustment cannot completely eliminate confounding, and randomized evidence remains necessary to establish comparative treatment effects.