Overall crude rates of local/regional recurrence (LRR) for patients with stage I patients vary from 3 to 11 %. This variation is due to a number of factors including the definition of “local recurrence” used, how local followed by distant recurrences are scored, the extent of surgical resection, and the ability to differentiate between a LRR and the development of a second primary lung cancer. Crude rates of local/regional first recurrence of 2.9–6.0 % and estimated 5-year LRR rates of 16–23 % have been reported in studies that define LRR as disease recurrence at the surgical margin, ipsilateral hilum, and/or mediastinum and report on patient cohorts undergoing curative lobar resection or greater without adjuvant therapy [1–4]. Whereas crude rates of 8–11 % and estimated 5-year LRR rates of 10–29 % have been reported using an expanding definition of LRR to include either the supraclavicular or contralateral lymph nodes [2, 5, 6]. Mean disease free intervals of 14.1–19.8 months for local/regional recurrences are similar to those reported for distant metastatic spread [7, 8]. Nearly 80 % of local recurrences occur in the first 2 years [9]. However, two distinct recurrence peaks occur at around 9 and 50 months post-treatment, with a smaller peak occurring around 30 months [10].

The risk of LRR can be influenced by surgical, pathological, and patient factors. The extent of parenchymal resection for early stage lung cancer has an impact on the risk of LRR. The North American Lung Cancer Study Group randomized 276 patients with T1N0 NSCLC to lobar versus adequate sublobar resection (wedge resection or segmentectomy with ≥ 2 cm margins) [3]. Of 247 patients eligible for analysis, 11 % (28/247) of patients were reported to have a LRR. However, patients undergoing sublobar resection had a significantly greater rate of LRR than those undergoing lobar resection (17 % vs. 6 %, p = 0.008). In addition, the effect of sublobar resection on LRR rates applied regardless of the type (wedge or segmentectomy). However, locoregional recurrence rates were higher, possibly as a result of a substantial proportion of large tumors (2–3 cm). Non-randomized trials demonstrate lower locoregional recurrence rates with segmentectomy compared to wedge resection for tumors ≤2 cm. Inadequate nodal assessment (<15 lymph nodes) has been associated with increased local recurrence rates in two prior studies, presumably based on incomplete staging[11, 12]. A third study did not find that the number of resected N1 or N2 nodes was associated with local failure [1]. However, the median number of resected N1 and N2 nodes in their report was two for each station and may have thus confounded their results.

A number of pathologic factors have also been associated with increased risk for LRR. Both increasing size (hazard ratio 2.0) and vascular invasion (HR 2.5) have been reported to be independent predictors of both distant and LRR in stage I patients [1, 7].

Patient factors can also contribute to the risk of LRR. Diabetes (HR 1.8) has been demonstrated to be an independent predictor of LRR in a single report that used an expanded definition of LRR [1]. While a history of smoking in of itself has not been identified as a predictor of LRR, a >20 pack years smoking history has been associated with an increased risk of LRR [12].

There have been four studies that have reported on postrecurrence survival (PRS) for patients with stage I disease [9, 13–15]. Shimada et al. followed 919 patients to determine factors influencing postrecurrence survival (PRS) and the effect of postrecurrence therapy (PRT) on PRS [13]. Type of recurrence included only local recurrence in 43 patients (25.3 %), distant in 113 (66.5 %), and both in 14 (8.2 %). Of the patients experiencing recurrence, 118 patients (69.4 %) received some form of treatment. However, only 10 % (4/43) with local only recurrence underwent a second intra-thoracic operation. The 1- and 2-year PRS proportions were 73.5 and 51.4 %, respectively. Multivariate analysis demonstrated that receiving any type of PRT was associated with improved PRS; however, being treated with surgery was not. In addition, mode of presentation was not a covariate included in the study multivariate analysis. Nagakawa et al. also evaluated the clinical outcomes of patients with resected lung cancer for postrecurrence prognostic factors [15]. Overall, 22 % (87/397) of patients had a recurrence, of whom 86 % (75/87) had some form of treatment. Reported 1-year and 3-year survival rates were 67.7 and 34.4 %, respectively. There was no significant difference in survival in those patients with local only vs. distant metastases or in those patients that were treated surgically or non-surgically. Once again, receiving some form of treatment was an independent predictor of improved PRS on multivariate analysis. Another study evaluated the prognostic predictors of PRS in patients with local only or local and distant recurrence [9]. Post-recurrence survival in patients with local only recurrence was not significantly different from that in those with both local and distant recurrences. Only treatment for initial recurrence was a significant predictor of post-recurrence survival in multivariate analyses.

In summary, receiving any form of treatment rather than palliation has been identified as a predictor of improved PRS regardless of the pattern of recurrence. However, undergoing surgery has not been demonstrated to improve survival on multivariate analysis, even when comparing surgical vs. non-surgical treatment for LRR. This is likely contributed to by the high re-recurrence rate after local therapies and that LRR may be followed by a distant recurrence in up to 20 % of patients [16]. Therefore receiving systemic rather than local therapy may be of more importance even when considering LRR.

Mode of presentation has been demonstrated in some studies to be a predictor of poor recurrence free survival. This is thought to be as a result of a high association of symptomatic recurrence with distant metastases which would preclude possible curative re-resection. Scheduling frequent follow-up imaging could theoretically identify isolated asymptomatic LRR at a time when curative intent treatment could be provided, however, there is no evidence to support this.

Lamont et al. followed 124 patients using a systematic post-operative surveillance protocol employing CT and CXR. In total 7 % (9/124) developed an isolated local recurrence. Although 89 % (8/9) of isolated local recurrences were asymptomatic at the time of detection, only one of nine was able to be resected [17]. Another study divided patients with resected NSCLC into “intensely” followed versus “non-intensely” followed based on a number of arbitrary criteria [18]. The authors were unable to show any differences between the groups with regard to detection of local recurrences or overall survival. Disease-free and median survival were similar between two groups of curatively resected NSCLC patients followed using a protocol resembling current National Comprehensive Cancer Network (NCCN) guidelines versus follow-up on a symptom oriented basis [19]. Walsh et al. investigated the development of relapse or second primary tumor in patients who experienced recurrence to identify predictors of decreased survival; a disease free interval of ≤12 months was the most significant predictor of decreased survival [20]. Furthermore, when controlling for a disease free interval >12 months, there was no difference in survival for those patients who presented with or without symptoms or based on whether they were treated with palliative or curative intent. This suggests that tumor biology, rather than mode of detection, may be the most important predictor of survival. Westeel et al. prospectively followed patients for the development of recurrence [21]. Compliance was high with scheduled diagnostic procedures/imaging (83–93 %). Univariate analysis demonstrated that disease free interval ≤12 months and symptomatic mode of presentation were associated with decreased survival. However, fiberoptic bronchoscopy accounted for 33 % of asymptomic intra-thoracic recurrences amenable to curative treatment, and about 60 % of their patients underwent pneumonectomy. Resection with pneumonectomy could alter locoregional recurrence rates and exaggerate the importance of bronchoscopy in postresection surveillance. All of the preceding studies, however, have included patient cohorts with both early and late stage disease.