Patterns of Failure after Resection of Non-small-cell lung cancer: Implications for Postoperative Radiation Therapy Volumes

Autor: Reviewer: Chika Madu, MD
Contribuidor de contenido: Abramson Cancer Center of the University of Pennsylvania
Fecha de la última revisión:

Authors: Kelsey CR, Light KL, Marks LB
Reference: Int J Radiat Oncol Biol Phys. 2006 Jul 15;65(4):1097-105
Affiliation: Department of Radiation Oncology, Duke University Medical Center


  • Several studies have shown that local recurrence rates for early stage non-small cell lung cancer (NSCLC) can approach as high as 20%.
  • For advanced disease, the recurrence rate can approach 40-50%.
  • One way to decrease local recurrence rates may be to predict the most likely sites of recurrence and prophylactically treat these areas with radiotherapy.
  • The problem with this concept is defining the indications for treatment and the sites that should be treated.
  • Patterns of relapse have been investigated in the past using techniques such as lymphangiograms.
  • The goal of this study is to analyze locoregional patterns of failure after resection of NSCLC.


  • This was a retrospective analysis of 61 patients who had resection of their NSCLCs at Duke University Medical Center between 1995 and 2005.
  • Inclusion criteria include: margin negative resection of all gross disease, no neoadjuvant or adjuvant RT, involvement of a local/regional site with or without metastasis at time of first recurrence, and availability of imaging studies (CT +/- PET) obtained at the time of relapse.
  • Patients were excluded if they met none of the above criteria, or if they had a simultaneous or sequential lung cancer.
  • Operative notes and pathology reports were reviewed for: location, size, histology, stage, margins, node sampling, and neoadjuvant or adjuvant chemotherapy data.
  • Imaging studies were reviewed by 1 of 2 authors (CK or LM).
  • Scoring: tumor recurrence at the bronchial stump after lobectomy/pneumonectomy or at the staple line after wedge resection was scored as a stump recurrence.
  • Mediastinal relapse was scored according to the AJCC staging nomenclature for LN stations.
  • 1 cm on CT scan was used as the threshold size for scoring new/enlarging hilar/mediastinal nodes.
  • All PET-positive nodes or soft tissue were scored regardless of size.
  • Bronchoscopy/mediastinoscopy findings were used to supplement imaging findings.
  • Diagrams illustrating failure sites for all patients were created.
  • Also created were summation diagrams illustrating sites of failure based on primary tumor lobe.
  • A series of theoretical AP/PA fields were created from the CT data set of a single patient with a right upper lobe tumor.
  • Each site of potential relapse was defined, and around this a series of enlarging radiation fields were plotted (variable clinical target volumes, or CTVs, plus a 1.5 cm margin)
  • A treatment plan was generated for each scenario, utilizing a dose of 45 Gy in 1.8 Gy fractions
  • For each field, the authors determined the V 20 (which is the volume of lung getting 20% of the total radiation dose), as well as the percentage of patients in whom all sites of locoregional relapse would have been covered by the radiation fields.

**Note: R= right; L= left; UL= upper lobe; LL= lower lobe


  • Most tumors were in the RUL and LUL, and most patients had a lobectomy and LN sampling at the time of surgery.
  • About 70% had stage 1 disease and 21% had stage III disease.
  • Median interval from surgery to relapse was 12 months (range 1.5-122 months).
  • About 44% had no distant metastases at time of local recurrence.
  • There was pathological confirmation of relapse in only 61% of patients.
  • For all patients, the most common site of failure was the stump/staple line [44% overall - wedge (79%) > lobectomy/pneumonectomy (34%), p=sig]
  • There was no difference in the rate of mediastinal relapse with regards to initial LN sampling, or in those with N1/N2 vs. N0 disease.
  • The supraclavicular fossa was more often involved in node (+) disease (27% vs. 4%, p=sig).
  • With regard to sites of relapse, the patterns are as follows: (numbers represent lymph node levels)
    • RUL: (26) most common site was 4R>stump>7>2R
    • RML: (2) 1 at stump, 1 with 4R, 2R, 7, and supraclav
    • RLL: (6) ipsilateral hilum, 4R
    • LUL: (20) stump> ipsilateral hilum>contralateral paratracheal> ipsilateral paratracheal
    • LLL: (7) 7>contralateral paratracheal>ipsilateral paratracheal
  • V20 increased with increasing volume of the simulated treatment fields, as one might expect.

Authors' Conclusions

  • The goal of this study was to better understand the patterns of spread of lung cancers - this can be done anatomically or clinically.
  • Based on anatomic studies, it is possible for tumors to bypass intrapulmonary LN and hilar LN to involve the mediastinal nodes; this occurs 38% of the time in upper lobe primary tumors and 20% of the time in lower lobe tumors.
  • In this clinical study, the mediastinal relapse without evidence of hilar relapse was 59%.
  • It is possible to have contralateral mediastinal and supraclavicular spread, especially with left-sided primary tumors.
  • Right-sided tumors and left-sided tumors drain to ipsilateral paratracheal nodes the majority of the time.
  • Subcarinal nodes are with lower lobe tumors more often than upper lobe tumors.
  • The surgical stump is the most common site of relapse, especially with wedge resections.
  • Locoregional relapse of NSCLC follows a fairly predictable pattern after resection on the basis of the originally involved lobe - this data may help clinicians construct PORT volumes that are smaller than ones traditionally utilized, potentially decreasing treatment-related toxicity.

Reviewer's Comments

Postoperative RT can improve local control in NSCLC; however, RT-associated toxicity poses some limitations. With more modern techniques, mediastinal radiation is better tolerated. This is in contrast to the PORT meta-analysis, in which no clear benefit was found except in subset analyses. This was likely due to older techniques and poor quality control. PORT in NSCLC continues to be utilized in clinical practice today. The question remains, however: what is an appropriate PORT field? Inclusion of all sites of possible mediastinal relapse is likely to be associated with excess toxicity. An alternative approach would be to treat the sites at highest risk for relapse, minimizing normal tissue toxicity, and potentially improving the therapeutic ratio. As systemic therapy improves, locoregional control will become even more important. This study helps define the likely patterns of relapse in NSCLC and may serve as a guide when designing optimal PORT fields.


Thank you for your feedback!