Colorectal cancer liver metastases – a population-based study on incidence, management and survival

Colorectal cancer (CRC) is a leading cause of cancer-associated death in Western populations and the third most frequent cause of cancer-related death in the world [1]. Population-based studies have shown that around 25–30% of patients diagnosed with CRC develop liver metastases during the course of their disease [2, 3]. Indications for curative-intended treatment of CRC liver metastases (CRCLM) have expanded in recent years. Unfortunately, despite the oncological and surgical advances made, only about 25% of patients affected are amenable to resection, which is regarded as the only way to achieve cure [3]. Historically, the indication for resection of liver metastases was based on tumour-related factors, for example tumour number, size and distribution in the liver. Currently the focus is rather on the future liver remnant (FLR), with resectability defined as the ability to perform a complete (R0) resection, while preserving a sufficient FLR. The presence of unresectable extra-hepatic disease is still considered a contraindication to liver surgery [4]. Liver resection can achieve 5-year survival rates of above 50%, compared to only around 5% for patients treated with palliative intent [5].

Although results are not consistent, primary tumour location in terms of right- versus left-sided cancer seems to play a role in metastatic pattern and survival [6‐9]. The observed differences in survival may depend on differences in embryologic origin, faecal exposure of the bowel, molecular profile, response to chemotherapy as well as the difference in time of detection with right-sided cancer generally presenting at a more advanced stage [7, 10, 11]. A number of studies on differences between right- and left-sided colon cancers that only included patients with resected stage I-III colon cancer found no difference in survival, or even improved survival for right-sided colon cancer [12, 13]. Studies on stage IV CRC showed a higher incidence of liver and lung metastases in left-sided colon cancer [6, 7]. Since right-sided metastatic cancer still implies a worse survival, it has been speculated whether the delay in diagnosis for right-sided cancer results in more extensive metastatic disease at diagnosis [7]. If so, it could explain the lower resection rate of liver metastases from right-sided cancer reported in some studies [6, 14].

The aim of the study was to describe the liver metastatic patterns and survival in a population-based cohort as a function of primary tumour characteristics and different combinations of metastatic disease.

In this study, the incidence of CRCLM was lower than the 50% often cited in the literature, but similar to results from previously reported population-based studies [2, 3]. Liver metastases were more often detected in the lower age groups as compared to the older age groups, which could possibly be attributed to a lower tendency to perform an extensive liver work-up in the old and frail. CRC originating from the left colon and rectum had a higher incidence of liver metastases compared with that of right-sided cancer. In spite of that, patients diagnosed with liver metastases secondary to a right-sided colon cancer had a higher T- and N-stage stage at diagnosis. Previously published retrospective analyses have shown the same pattern, with right-sided colon cancer having a higher TNM stage at diagnosis [11, 16] and lower incidences of liver metastases [6, 10]. Price et al. speculated on whether the delay in diagnosis for right-sided cancer could result in more extensive metastatic disease at diagnosis, explaining the worse survival in metastatic right-sided cancer [7]. The present study supports this hypothesis with a higher number of metastases and more liver segments involved at detection, observed in patients with liver metastases from a right-sided colon cancer. Since most other studies present the presence of metastatic disease as a dichotomised value (yes or no), the information regarding differences in the actual extent of the disease is usually scant. Lung metastases, typically detected a year later than liver metastases, were also significantly more often detected in patients with left-sided colon and rectal cancer, compared to patients with right-sided cancer, while peritoneal dissemination was more often seen in right-sided cancer. There are known molecular differences between right- and left-sided colon cancer with the former more often being poorly differentiated, as well as more often KRAS and/or BRAF mutated [8, 10, 17]. RAS mutations have been reported to be associated with a more aggressive tumour biology and occur in 35–45% of all patients with metastatic CRC [18, 19]. Among patients with resectable liver metastases, the presence of RAS mutations is 10–15% lower, indicating that the underlying biology is prognostic and influences the likelihood of surgical resection [20]. Furthermore, patients with RAS mutations undergoing liver resection have worse overall and recurrence-free survival [20].

Since this study included reviews of abdominal imaging, detailed information of intrahepatic metastatic spread can be presented. Almost one-third of all patients with metastases in the liver had widespread disease at diagnosis, engaging all segments. On the other hand, nearly 50% had liver-only metastases at presentation with metastatic disease, indicating a more pronounced intrahepatic spread than previously reported [2, 3]. This might in part be explained by the true population-based nature of this study, where patients were not excluded from analysis based on treatment status of the primary tumour.

Patients with liver metastases from a right-sided cancer had a significantly worse overall survival compared to patients with liver-metastasized left-sided colon cancer. This is in agreement with previous reports of inferior survival of right-sided colon cancer in the presence of metastatic disease [7, 14]. Norén et al. also reported better survival in liver metastatic left-sided CRC and further demonstrated that the resection rate of liver metastases varied according to site of the primary tumour, with lower resection rates for right-sided cancers [6]. This was also found to be true in the present study, where patients with liver metastatic right-sided cancer were resected for their liver metastases less often and had an intermediate survival as compared to patients resected for liver metastases from a left-sided cancer and patients not resected at all. However, tumour site was not significantly associated with the likelihood of undergoing a resection in the multivariate logistic regression analysis. Reasons for these survival differences are most likely multifactorial. Patients with right-sided cancer present later, are older at diagnosis and may have more comorbidities than patients with left-sided cancer [11, 21]. In a study by Gervaz et al. the difference in stage could not explain survival differences, since the survival difference was still present when adjusting for multiple factors, including stage [11]. The stage-adjusted subgroup analysis in this study was limited by the low number of patients in each stage group but still, the results were similar to other studies, with stage III and IV right-sided cancer having significantly worse survival compared with that of left-sided cancer [8, 12]. Stage II right-sided cancer on the other hand is associated with better survival compared with that of left-sided [12, 13, 22]. This might be explained by a higher proportion of right-sided stage II cancers having microsatellite instability, which in turn is associated with a more favourable outcome and a decreased likelihood of distant organ metastases at diagnosis [23]. In a recent publication by Warschkow et al., using propensity score analysis on patients with resected stage I-III colon cancer, a worse survival for left-sided colon cancer was found which contradicts the previous paradigm [13]. A Danish population-based study challenged the right/left categorization since no clear trend was found regarding survival and their more detailed colon sub-site analysis revealed a much more complex picture [9].

The proximal and distal colon segments are of different embryological origins, where the former develops from the midgut and the latter from the hindgut. Most other studies on location of colon cancer as a prognostic factor for survival do not include rectal cancer. In the few studies that do, the results are consistent with right-sided cancer having worse survival compared with left-sided cancer [7, 10] When analysing early stage disease, there could be merit in not including rectal cancer in in the analyses due to different treatment strategies, but in the setting of metastatic disease treatment of metastases is not different between metastatic rectal and colon cancer. Even when excluding rectal cancer from left-sided colon cancer in the subgroup analysis in this study, the findings remained comparable with primary tumour site still being a statistically significant prognostic factor, influencing survival.

The location of extra-hepatic disease is known to affect survival, with lung metastases having better outcomes [24]. The present study shows that patients with metastases confined to the lungs did much better and had three times longer median survival (4.3 years) compared to patients with liver-only metastases. In the multivariate analysis of factors influencing survival, the presence of lung metastases did not significantly influence overall survival in patients with liver metastases, nor in the overall patient population. This finding could stimulate the debate on whether there is merit in treating liver metastases in a subset of patients with unresectable lung metastases. Such an approach is also supported by data reported by Dave et al. showing a 30% 5-year survival in a cohort of patients that were planned for liver and lung resection, but where the pulmonary treatment was not carried out due to tumour progression after liver resection [25]. The expected 5-year survival in these patients were much less, indicating that resection of the liver metastases had an influence on survival.

In the present study the referral rate to a liver MDT was higher than the figure reported by Young et al. where 22.5% of patients with CRCLM were referred for evaluation by a liver surgeon [26]. As a likely consequence, the resection rate of liver metastases in this study (25.4%) were higher than the resection rate in the Young study (17%) [26], but in the same range as other population-based studies [3]. The 5-year survival rate of 48.6% in patients resected for liver metastases presented in this study indicates that high survival rates can be achieved even in a population-based setting, where referral and selection bias could be assumed as being minimal. It is clear that with applying current indications for curative-intended treatment of CRCLM, not all resected patients will benefit from the procedure in terms of OS. Conversely, some patients with liver metastases that currently are assessed as not suitable for curative-intended treatment might have a survival benefit from hepatic disease-control.

A weakness of the present study is the retrospective design where follow-up of patients could have been influenced by differences in adherence to guidelines by the treating physicians. Some patients could have had a less stringent follow-up because of advanced age, severe comorbidities or early carcinomas perceived to have a lower risk for metastasizing. Even if it is unlikely to have affected overall survival, it could to some extent have affected the time of detection of metastases. A major limitation of this study is the lack of knowledge of the American Society of Anaesthesiologists (ASA) classification that could potentially be a confounder in the survival analyses. Due to the limited number of patients in the group of metastatic right-sided colon cancer, a proper stage-adjusted analysis could not be conducted. This of course makes inference on survival in these groups uncertain. Lung metastases did not significantly influence survival in the multivariate Cox regression analysis, but one cannot exclude immortality bias, since lung metastases were found to be diagnosed later in the course of the disease. Furthermore, no information on tumour markers and mutation status of the primary tumours was available, which could possibly shed light on the difference in survival seen between liver metastatic right-sided and left-sided colon cancer.

In conclusion, the study offers detailed population-based data on the metastatic pattern of CRC that could assist in more structured and individualized guidelines for follow-up and treatment of CRCLM. It also supports earlier findings regarding metastatic CRC, namely that right-sided tumours are associated with worse outcomes compared to left-sided tumours. Moreover, data suggest that lung metastases are not always associated with poor survival. With increasing focus on the benefit of liver intervention in terms of prolonged survival, there might be a subgroup of patients that could benefit from liver intervention, despite the presence of lung metastases not suitable for treatment.