This study focused on the effects of peri-operative chemotherapy in the setting of pulmonary resection for colorectal lung metastases. We believe this is the first dedicated published series on peri-operative chemotherapy. The absolute 5-year survival rate was higher in the CS group compared with S group (74% versus 68%) as well as 5-year RFS (38% versus ≤18%) despite a shorter DFI. Lymph node involvement was the only significant prognostic factor found in our cohort but only in univariate analysis. The numbers were too small to compare lymph node involvement between the perioperative chemo vs no chemo groups or to perform a multivariate analysis. Also, the more recent procedures almost routinely involved lymph node resection whereas earlier procedures did not, which could be impacted on by numerous confounding factors. Toxicity of chemotherapy was manageable with no post-operative deaths and only a 4% surgical complication rate observed in the chemotherapy group. Neoadjuvant chemotherapy and targeted agents did not cause any evidence of pulmonary toxicity, either radiologically or histologically. FDG-PET scanning, also specifically evaluated for the first time in our series, provides higher concordance with the number of histologically proven metastases present than CT when PET-avid lesions are present (72% versus 61%).
Published series of pulmonary metastasectomy have included variable numbers of patients who received systemic peri-operative treatment but detailed outcomes of these subgroups have not been specifically reported. In a systematic review of 11 retrospective studies including 1307 patients , 8 of the 11 studies reported the use of neoadjuvant or adjuvant chemotherapy but discussion of toxicity, response, details of the regimen were consistently lacking. Two reports within the review, and two further published series have evaluated peri-operative chemotherapy as a prognostic factor. Saito et al . (n = 165) and Lee et al . (n = 59) both found administration of peri-operative chemotherapy to be non-significant as a prognosticator in multivariate analysis. In a series of 30 patients, 5-year OS appeared higher in patients receiving post-operative chemotherapy (83.5 versus 57.1%) yet 5-year disease-free survival was similar (45.8% versus 46.9%), though in this small cohort, neither was statistically significant (p = 0.397 and 0.754 respectively) . A larger study of 315 patients by Kanemitsu et al . reported 27.2% receiving ‘adjuvant chemoradiotherapy’. Univariate analysis demonstrated a significant benefit in undergoing adjuvant therapy (Hazard ratio for death at 3 years 0.69, 95% CI 0.49-0.98; p = 0.037) with a non-significant trend towards reduction in death at 3 years in multivariate analysis (HR 0.71, 95%CI 0.49-1.03; p = 0.068).
The 5-year OS of 72% in our cohort was higher than other published series. The recent systematic review reported a median 5-year OS of only 39.6% (range: 24-56%) for R0 resected patients . Our OS was promising and the low morbidity and mortality we reported supports the use of pulmonary resection and peri-operative chemotherapy as a treatment modality. With our 5-year RFS of 31% after initial resection and 3 additional patients being rendered disease-free after subsequent resections, pulmonary resection appears beneficial in this selected group, and re-resection in a small number of patients is also a feasible option.
Neoadjuvant chemotherapy also has the advantage of assessing in vivo tumour response and overall course of the disease as well as potentially downsizing lesions. Whilst we acknowledge that the majority of lung lesions in this study are not evaluable by RECIST due to small size, a reduction in radiological tumour size was documented in 62% of resections where neoadjuvant chemotherapy was administered, with an additional 30% having stable disease. Indeed, although the presence of necrosis and, to a lesser extent, fibrosis are features seen in metastatic colorectal carcinoma, the extent of these changes being up to 80% likely reflects partial response to chemotherapy prior to surgery in some cases. Progression was seen in only a small minority (8%) during neoadjuvant therapy. Obviously, due to selection bias, we recognise that the cohort of patients who progressed, becoming unresectable, were not captured in this study nor did any lesions require downsizing prior to consideration of resection. However, this response was similar to published response rates in metastatic CRC of 36-66% [22–24].
DFI following curative resection of the primary has been described as a possible prognostic determinant. In the CS group, DFI was shorter than in the S group, suggesting that those with a shorter DFI are being treated more aggressively. Despite the slightly shorter DFI in this group (20.5 months versus 27.5 months, p = 0.277), the absolute 5-year OS was higher (74% vs. 68%).
This study demonstrates that chemotherapy toxicity appears manageable in this population. Pfannschmidt’s 2010 systematic review reported a 0–2.4% post-operative mortality, consistent with our observation of no surgery-related deaths . Acute surgical complications were observed less in the chemotherapy group in our study (3 events in 68 resections versus 5 events in 16 resections for surgery alone). Neoadjuvant chemotherapy was better tolerated than adjuvant chemotherapy with fewer recorded toxicities (reported in 27% versus 54% of patients). This difference in toxicity pre- and post-surgery is consistent with reports in operable rectal and gastric cancers where neoadjuvant and adjuvant chemotherapy have been evaluated [25–27]. Of interest, in patients who received targeted therapy, only 2 complications occurred. A patient receiving neoadjuvant sunitinib plus FOLFIRI (within the context of a phase I dose-escalation study) developed grade 4 neutropenia pre-operatively and a patient receiving FOLFIRI chemotherapy plus cetuximab had a prolonged air-leak. None of the toxicities or surgical complications occurred in the 6 patients receiving chemotherapy plus bevacizumab with the median interval between end of neoadjuvant treatment and surgery of 37 days, despite the widely published wound-healing, bleeding and thromboembolic complications known to occur with bevacizumab. There was no radiological evidence of chemotherapy-related pulmonary toxicity and histological evaluation of the resected tissue demonstrated no evidence of toxic changes to non-tumour lung parenchyma. This is reassuring, and certainly in contrast to that of neoadjuvant chemotherapy in the setting of liver resection, where it is well documented that irinotecan can cause steatohepatitis and oxaliplatin is associated with sinusoidal obstruction syndrome . There are currently no data on chemotherapy-induced lung toxicity seen histologically within the context of resectable CRC pulmonary metastases and our report represents first preliminary evidence demonstrating no significant histological damage is detected following CRC chemotherapy.
CT is not highly accurate in differentiating malignant from benign pulmonary nodules, hence many benign lesions are removed in pulmonary metastasectomy. FDG-PET is used routinely to confirm a single metastatic site of disease prior to resection however may assist in further delineating malignant involvement of pulmonary lesions. Concordance of FDG-PET with histological involvement of pulmonary nodules has not previously been assessed. In patients undergoing neoadjuvant chemotherapy, the timing of preoperative PET varied in our cohort, yet administration of chemotherapy did not impact on FDG-avidity of the lesions. In PET-positive patients, we observed a higher concordance between number of lesions with positive histology and PET than CT. These concordance rates were exploratory and hence only described in absolute values, the limited sample size precluded any formal statistical testing. With further evaluation, this may have important consequences on the decision of which lesions to resect at surgery in addition to excluding extrapulmonary disease.
The most important limitation of our study was the selectivity of our approach. Included patients needed to have resectable pulmonary metastases, physical fitness for lung resection and no progression during neoadjuvant chemotherapy to the extent of being denied surgery. In patients who had significant co-morbidities, radiofrequency ablation might have been utilised as an alternative. High usage of PET scanning in our study ensured that patients with extra-thoracic non-hepatic metastases were excluded from pulmonary resection. As an indication, our institution commenced 670 patients on first line chemotherapy for metastatic CRC between 2000 and 2008 (equivalent of about 80 patients per year) .