National variation in pulmonary metastasectomy for colorectal cancer

Abstract Aim Evidence on patterns of use of pulmonary metastasectomy in colorectal cancer patients is limited. This population‐based study aims to investigate the use of pulmonary metastasectomy in the colorectal cancer population across the English National Health Service (NHS) and quantify the extent of any variations in practice and outcome. Methods All adults who underwent a major resection for colorectal cancer in an NHS hospital between 2005 and 2013 were identified in the COloRECTal cancer data Repository (CORECT‐R). All inpatient episodes corresponding to pulmonary metastasectomy, occurring within 3 years of the initial colorectal resection, were identified. Multi‐level logistic regression was used to determine patient and organizational factors associated with the use of pulmonary metastasectomy for colorectal cancer, and Kaplan–Meier and Cox models were used to assess survival following pulmonary metastasectomy. Results In all, 173 354 individuals had a major colorectal resection over the study period, with 3434 (2.0%) undergoing pulmonary resection within 3 years. The frequency of pulmonary metastasectomy increased from 1.2% of patients undergoing major colorectal resection in 2005 to 2.3% in 2013. Significant variation was observed across hospital providers in the risk‐adjusted rates of pulmonary metastasectomy (0.0%–6.8% of patients). Overall 5‐year survival following pulmonary resection was 50.8%, with 30‐day and 90‐day mortality of 0.6% and 1.2% respectively. Conclusions This study shows significant variation in the rates of pulmonary metastasectomy for colorectal cancer across the English NHS.

lished guidance on the management of metastatic disease in colorectal cancer in January 2020. They found no randomized trials and based their guidance on a 'very low quality' follow-up analysis [6] but still recommended that lung metastasectomy should be 'considered' [7]. Finally, the Society of Thoracic Surgeons Consensus Document on Pulmonary Metastasectomy in 2019 [8] again, despite acknowledging the lack of randomized evidence, gave as their first recommendation that pulmonary metastasectomy should be 'considered' and 'carefully individualized' [8].
'Pulmonary metastasectomy in colorectal cancer', published in 2020 [9,10], was the first randomized trial on this topic. Patients were randomized to lung metastasectomy (N = 46) or control (N = 47) and 87% were followed for more than 5 years or until death. There was no significant difference comparing metastasectomy to controls; the hazard ratio (HR) for death was 0.93 (95% CI 0.56, 1.56). This puts into question the recommendations for practice from ACPGBI, Society of Thoracic Surgeons and NICE.
It had been estimated that 15%-25% of those with pulmonary metastases were being considered for local treatments [11].
Traditionally surgery has been the main local treatment, but more recently image guided thermal ablation (IGTA), including radiofrequency, microwave or cryoablation, and stereotactic radiotherapy (variously abbreviated as SABR or SBRT) have been increasingly used as alternatives. Each has been tested in a randomized controlled trial [12,13] but, unfortunately, both had major imbalances across trial arms so, again, it is not clear what constitutes gold standard care.
Further studies with an emphasis on understanding the extent and characteristics of patients with metastases are required.
Arguably more information is required about the disease in its totality rather than outcomes amongst patients carefully selected for local treatments. Most existing studies on pulmonary metastasectomy are single-centre surgical case series [8] with very few population-based cohorts or registry studies [14][15][16][17]. When the larger series within these studies are considered, the incidence of pulmonary metastasectomy is estimated to be only around 2.5% [14,16] of all colorectal cancer patients and 3.5%-5% [14,17] of those with metastases at diagnosis. As such, pulmonary metastasectomy is clearly used in a highly selectively manner.
This population-based study therefore aimed to investigate the use of pulmonary metastasectomy in the colorectal cancer population across the English NHS and to quantify the extent of any variations in practice and outcome.

ME THODS
All adults diagnosed with a first primary colorectal cancer (ICD-10 codes C18-C20), and who had undergone a major resection for their disease in an NHS hospital with a multidisciplinary team   Table S1). The lung resection codes were presumed to correspond to pulmonary metastasectomy. It is possible that some of these procedures may have been carried out for primary lung cancer or other diagnoses, due to the absence of additional information. Primary tumours of the caecum, appendix, ascending colon, hepatic flexure and transverse colon (ICD-10 codes C18.0-C18.4) were assigned as right-sided colon tumours, whilst tumours in the splenic flexure, descending colon, sigmoid colon and rectosigmoid (ICD-10 C18.5-C19) were assigned as left-sided colon tumours. Rectal tumours were assigned

What does this paper add to the literature?
There is limited population-based evidence available on patterns of pulmonary metastasectomy for colorectal cancer lung metastases. This is one of the first studies to use large population-level, linked datasets to investigate practice and outcomes following such surgery across the English National Health Service.
using ICD-10 C20. Where more than one OPCS procedure code appeared on the same day for the lung resection, the most extensive operation was selected.
The Charlson Comorbidity Index (CCI) [18] was derived for each patient, taking into account diagnoses (excluding cancer) from any hospital admissions in the year preceding diagnosis of colorectal cancer. The cancer component of the CCI was derived from the cancer registry information in CORECT-R and added to that obtained from HES data. The CCI was categorized as 0, 1, 2 and ≥3 with higher scores indicating greater degree of comorbidity.
Data from the Organizational Survey 2016, carried out by the National Bowel Cancer Audit [19] were used to identify whether the initial colorectal resection was carried out within an MDT with an on-site specialist thoracic team. Twenty-five specialist thoracic surgery centres were identified in the data out of 146 Trusts.
Multi-level logistic regression was used to determine factors associated with the use of resection for lung metastases. Models were constructed with patients clustered within hospitals, which were then further clustered within Cancer Alliances. Explanatory variables in the risk-adjusted model were age at resection, sex, IMD quintile, tumour site, year of primary major colorectal resection, CCI, stage at diagnosis, and whether the trust was a thoracic surgery centre. Funnel plots were constructed to show the variation in resection of metastases across MDTs using the Spiegelhalter approach [20] and those MDTs outside the 99.8% control limits were considered 'outliers' in terms of their practice.
Survival was calculated from the date of lung metastasectomy until death, or censored on 31 December 2016. Overall survival following pulmonary metastasectomy was analysed using Kaplan-Meier actuarial methods by sex, age at diagnosis, IMD quintile, site and stage of primary tumour at diagnosis and CCI. Log rank tests were used to test for any statistically significant difference between these groups. Cox proportional hazards regression analysis was used to determine factors associated with a higher risk of death following pulmonary metastasectomy surgery. Analyses were conducted using STATA 15 (StataCorp).
The best estimate of the representative median interval between primary colorectal cancer resection and pulmonary metastasectomy operations is 29 months, based on data from seven reports from 2006 to 2013 including a total of 1606 operations [21][22][23][24][25][26][27].
Our 3-year cut-off will have missed some cases so a sensitivity analysis was also carried out to identify pulmonary metastasectomies

Surgical management of metastases
During the period 1 January 2005 to 31 December 2013, 173 354 patients were identified as undergoing major resection for a colorectal tumour, the characteristics of whom are outlined in Table 1.
Of these 3434 (2.0%) underwent one or more resections for lung metastases within 3 years of their primary colorectal resection. ( Figure 1). The rate of pulmonary metastasectomy was higher for patients presenting with tumours in the rectum (3.17%) than for tumours of the colon (1.49%) which is a reflection of the well documented difference in incidence due to the anatomical difference in venous drainage between the colon and rectum [2,28,29].
Multi-level logistic regression was performed using data from 2010 to 2013, during which period staging of the primary tumour was more complete ( Table 2). The likelihood of receiving surgery reduced with age (OR 0.23 for age >80 compared to the average age group 61-70; 95% CI 0.18-0.29). There was no significant difference in the rate of surgery between men and women or across deprivation quintiles. The odds of resection decreased with increasing CCI (OR 0.44 for CCI 3 or more compared to a score of zero; 95% CI 0.27-0.71). Patients with rectal tumours were most likely to receive surgery (OR 2.61; 95% CI 2.30-2.97) and those with tumours in the right colon were the least likely to receive treatment for lung metastases. Patients with more advanced disease stage at diagnosis were more likely to undergo pulmonary metastasectomy within 3 years of their primary colorectal resection (OR for Stage IV vs. Stage III 1.77; 95% CI 1.55-2.03). These will include patients with synchronous metastases, which were already evident at the time of initial cancer staging. There was no statistically significant difference in the likelihood of pulmonary metastasectomy depending on whether patients received surgery for their colorectal primary in a trust with a specialist thoracic centre, in either the risk-adjusted or non-adjusted model.
There was a large degree of variation in the proportion of patients having a pulmonary metastasectomy when analysed by the hospital provider for their primary colorectal resection, with crude and risk-adjusted rates both between 0.0% and 6.8% (P < 0.001). By Cancer Alliance/Vanguard the crude rate varied between 1.5% and 3.1% (P = 0.003) and the risk-adjusted rate varied between 1.5% for Lancashire and Cumbria and 3.0% for South East London; however, there was no statistically significant difference between Cancer Alliances following risk adjustment ( Figure 2).

Survival
Overall median survival following pulmonary metastasectomy was  when there is no observed difference between the two procedures.
It is not possible to know how much this association was influenced by the severity of the disease or the treatment. There are few pneumonectomies and bilobectomies and survival was much worse for these patients compared to those undergoing lobectomy; again the disease and the treatment was more severe.
Potential volume effects on survival outcomes were considered; however, there was no statistically significant difference in 5-year survival following pulmonary metastasectomy between high, mid and low volume providers. There was no correlation between the crude or adjusted pulmonary resection rate and 5-year survival by Cancer Alliance or MDT.     This reflects the higher rate of metastasis to the lungs in rectal cancer, related to the anatomical differences in venous drainage already noted [2,28,29].

TA B L E 2 Odds of having a pulmonary metastasectomy within 3 years of primary colorectal tumour resection
Postoperative 30-and 90-day mortality of 0.6% and 1.2% respectively shows lung metastasectomy to be relatively safe. Overall Receiving surgery for the primary colorectal tumour, in a trust with a specialist thoracic surgical centre on-site, did not increase the chances of a patient receiving pulmonary resection in either the risk-adjusted or unadjusted model. This contrasts with liver metastasectomy whereby patients with colorectal cancer liver metastases are more likely to receive surgical resection when their primary colorectal tumour is resected in a hospital with an on-site hepatobiliary team [33]. This also contrasts with studies carried out on primary non-small-cell lung cancer, which suggest that being seen first in a specialist thoracic surgical centre increases the likelihood of receiving lung surgery [34].
There was significant variation in the percentage of patients receiving a pulmonary resection between trusts, with two trusts having higher than average pulmonary metastasectomy rates, that is, outside of the 99.8% confidence limits so outside of the 'random' variation that would be expected, and one trust having a lower than average pulmonary metastasectomy rate of only 0.14% of patients. Whilst there were significant differences in the crude rate of pulmonary metastasectomy by the Cancer Alliance within which the patient received their colorectal resection, this was no longer observed following risk adjustment for patient and tumour characteristics. Therefore it seems that referral practice is similar at a regional level, with more variation occurring on a hospital site level.
This variation may also be affected by trusts referring patients for ablation or SBRT rather than resection. Data on ablation were not fully captured during this work due to a large amount of variation in coding practice. Exploring this will form the basis for future work.
Current NICE draft guidance [7] suggests that there is insufficient evidence to recommend one type of local treatment over another so either surgical resection, ablation or SBRT should be considered for people with colorectal lung metastases. It will be interesting to note whether trusts with lower resection rates refer larger numbers of patients for ablation or SBRT and whether patients have similar access to all treatments across trusts.
The rate of pulmonary metastasectomy in the over 80s group was almost one-fifth of that for patients aged 71-80; however, there Unfortunately, we do not have information on the frequency of ablative treatments or radiotherapy in order to assess whether they receive alternative treatment.
There was no statistically significant difference in the likelihood of receiving a pulmonary metastasectomy across different deprivation quintiles, in contrast to the trend observed for resection for colorectal cancer liver metastases [35] or resection for primary lung cancer where more deprived patients are less likely to receive pulmonary resection [36,37].
One limitation of this study is that there is not sufficient information to ascertain the reason for the lung resection. It has been TA B L E 3 Cox proportional hazards model results for survival following pulmonary metastasectomy and survival following primary colorectal resection presumed that these surgical procedures are taking place due to pulmonary metastases, given the proximity to diagnosis of colorectal cancer; however, it is also possible that patients received surgery for other diagnoses such as primary lung cancer or emphysema.

CO N FLI C T O F I NTE R E S T S
None.