Differences in survival and recurrence of colorectal cancer by stage across population-based European registries.

Recurrence after colorectal cancer resection is rarely documented in the general population while a key clinical determinant for patient survival. We identified 8785 patients with colorectal cancer diagnosed between 2010 and 2013 and clinically followed up to 2020 in 15 cancer registries from seven European countries (Bulgaria, Switzerland, Germany, Estonia, France, Italy, and Spain). We estimated world age-standardized net survival using a flexible cumulative excess hazard model. Recurrence rates were calculated for patients with initially resected stage I, II, or III cancer in six countries, using the actuarial survival method. The proportion of nonmetastatic resected colorectal cancers varied from 58.6% to 78.5% according to countries. The overall 5-year net survival by country ranged between 60.8% and 74.5%. The absolute difference between the 5-year survival extremes was 12.8 points for stage II (Bulgaria vs Switzerland), 19.7 points for stage III (Bulgaria vs. Switzerland) and 14.8 points for Stage IV and unresected cases (Bulgaria vs. Switzerland or France). Five-year cumulative rate of recurrence among resected patients with stage I-III was 17.7%. As compared to the mean of the whole cohort, the risk of developing a recurrence did not differ between countries except a lower risk in Italy for both stage I/II and stage III cancers and a higher risk in Spain for stage III. Survival after colorectal cancer differed across the concerned European countries while there were slight differences in recurrence rates. Population-based collection of cancer recurrence information is crucial to enhance efforts for evidence-based management of colorectal cancer follow up.

A new cure model that corrects for increased risk of non-cancer death: analysis of reliability and robustness, and application to real-life data.

BACKGROUND: Non-cancer mortality in cancer patients may be higher than overall mortality in the general population due to a combination of factors, such as long-term adverse effects of treatments, and genetic, environmental or lifestyle-related factors. If so, conventional indicators may underestimate net survival and cure fraction. Our aim was to propose and evaluate a mixture cure survival model that takes into account the increased risk of non-cancer death for cancer patients. METHODS: We assessed the performance of a corrected mixture cure survival model derived from a conventional mixture cure model to estimate the cure fraction, the survival of uncured patients, and the increased risk of non-cancer death in two settings of net survival estimation, grouped life-table data and individual patients' data. We measured the model's performance in terms of bias, standard deviation of the estimates and coverage rate, using an extensive simulation study. This study included reliability assessments through violation of some of the model's assumptions. We also applied the models to colon cancer data from the FRANCIM network. RESULTS: When the assumptions were satisfied, the corrected cure model provided unbiased estimates of parameters expressing the increased risk of non-cancer death, the cure fraction, and net survival in uncured patients. No major difference was found when the model was applied to individual or grouped data. The absolute bias was < 1% for all parameters, while coverage ranged from 89 to 97%. When some of the assumptions were violated, parameter estimates appeared more robust when obtained from grouped than from individual data. As expected, the uncorrected cure model performed poorly and underestimated net survival and cure fractions in the simulation study. When applied to colon cancer real-life data, cure fractions estimated using the proposed model were higher than those in the conventional model, e.g. 5% higher in males at age 60 (57% vs. 52%). CONCLUSIONS: The present analysis supports the use of the corrected mixture cure model, with the inclusion of increased risk of non-cancer death for cancer patients to provide better estimates of indicators based on cancer survival. These are important to public health decision-making; they improve patients' awareness and facilitate their return to normal life.

Biliary tract cancers have distinct epidemiological patterns and clinical characteristics according to tumour site.

BACKGROUND: Little is known about the epidemiology of biliary tract cancers over the last decade. We investigated trends in incidence, treatment and prognosis of biliary tract cancers according to anatomic site. METHODS: 714 biliary tract cancers recorded between 2012 and 2019 in the French population-based cancer registry of Burgundy were included. Trends in world age-standardized incidence were depicted using Poisson regression. RESULTS: Intrahepatic cholangiocarcinoma accounted for 40% of biliary tract cancer. Half of the patients were older than 75 years at diagnosis. Incidence of biliary tract cancer did not vary over time, except a slight increase in intrahepatic cholangiocarcinoma in men and a decrease in the ampulla in both sexes. Among non-metastatic patients, the proportion who underwent R0 resection ranged from 15% for intrahepatic cholangiocarcinoma to 58% for ampulla cancer (p < 0.001). Age, performance status and hospital type were associated with resection. Among unresected patients, 45% received chemotherapy. Older age, jaundice, increasing performance status and comorbidities index negatively affected chemotherapy administration. Net survival was higher for ampulla than for other sites, regardless of patient and treatment characteristics. CONCLUSION: Biliary tract cancers present different patterns in incidence. The ampulla site should be considered separately in clinical trials due to its better outcomes.

Health status of prevalent cancer cases as measured by mortality dynamics (cancer vs. noncancer): Application to five major cancers sites.

BACKGROUND: Cancer prevalence is heterogeneous because it includes individuals who are undergoing initial treatment and those who are in remission, experiencing relapse, or cured. The proposed statistical approach describes the health status of this group by estimating the probabilities of death among prevalent cases. The application concerns colorectal, lung, breast, and prostate cancers and melanoma in France in 2017. METHODS: Excess mortality was used to estimate the probabilities of death from cancer and other causes. RESULTS: For the studied cancers, most deaths from cancer occurred during the first 5 years after diagnosis. The probability of death from cancer decreased with increasing time since diagnosis except for breast cancer, for which it remained relatively stable. The time beyond which the probability of death from cancer became lower than that from other causes depended on age and cancer site: for colorectal cancer, it was 6 years after diagnosis for women (7 years for men) aged 75-84 and 20 years for women (18 years for men) aged 45-54 years, whereas cancer was the major cause of death for women younger than 75 years whatever the time since diagnosis for breast and for all patients younger than 75 years for lung cancer. In contrast, deaths from other causes were more frequent in all the patients older than 75 years. Apart from breast cancer in women younger than 55 years and lung cancer in women older than 55 years and men older than 65 years, the probability of death from cancer among prevalent cases fell below 1%, with varying times since diagnosis. CONCLUSIONS: The authors' approach can be used to better describe the burden of cancer by estimating outcomes in prevalent cases.

Modeling excess hazard with time-to-cure as a parameter.

Cure models have been widely developed to estimate the cure fraction when some subjects never experience the event of interest. However, these models were rarely focused on the estimation of the time-to-cure, that is, the delay elapsed between the diagnosis and "the time from which cure is reached," an important indicator, for instance, to address the question of access to insurance or loans for subjects with personal history of cancer. We propose a new excess hazard regression model that includes the time-to-cure as a covariate-dependent parameter to be estimated. The model is written similarly to a Beta probability distribution function and is shown to be a particular case of the non-mixture cure models. Parameters are estimated through a maximum likelihood approach and simulation studies demonstrate good performance of the model. Illustrative applications to three cancer data sets are provided and some limitations as well as possible extensions of the model are discussed. The proposed model offers a simple and comprehensive way to estimate more accurately the time-to-cure.

Impact of time-varying cumulative bevacizumab exposures on survival: re-analysis of data from randomized clinical trial in patients with metastatic colo-rectal cancer.

BACKGROUND: As cancer treatment, biotherapies can be as effective as chemotherapy while reducing the risk of secondary effects, so that they can be taken over longer periods than conventional chemotherapy. Thus, some trials aimed at assessing the benefit of maintaining biotherapies during chemotherapy-free intervals (CFI). For example, the recent PRODIGE9 trial assessed the effect of maintaining bevacizumab during CFI in metastatic colorectal cancer (mCRC) patients. However, its analysis was hindered by a small difference of exposure to the treatment between the randomized groups and by a large proportion of early drop outs, leading to a potentially unbalanced distribution of confounding factors among the trial completers. To address these limitations, we re-analyzed the PRODIGE9 data to assess the effects of different exposure metrics on all-cause mortality of patients with mCRC using methods originally developed for observational studies. METHODS: To account for the actual patterns of drug use by individual patients and for possible cumulative effects, we used five alternative time-varying exposure metrics: (i) cumulative dose, (ii) quantiles of the cumulative dose, (iii) standardized cumulative dose, (iv) Theoretical Blood Concentration (TBC), and (v) Weighted Cumulative Exposure (WCE). The last two metrics account for the timing of drug use. Treatment effects were estimated using adjusted Hazard Ratio from multivariable Cox proportional hazards models. RESULTS: After excluding 112 patients who died during the induction period, we analyzed data on 382 patients, among whom 320 (83.8%) died. All time-varying exposures improved substantially the model's fit to data, relative to using only the time-invariant randomization group. All exposures indicated a protective effect for higher cumulative bevacizumab doses. The best-fitting WCE and TBC models accounted for both the cumulative effects and the different impact of doses taken at different times. CONCLUSIONS: All time-varying analyses, regardless of the exposure metric used, consistently suggested protective effects of higher cumulative bevacizumab doses. However, the results may partly reflect the presence of a confusion bias. Complementing the main ITT analysis of maintenance trials with an analysis of potential cumulative effects of treatment actually taken can provide new insights, but the results must be interpreted with caution because they do not benefit from the randomization. TRIAL REGISTRATION: clinicaltrials.gov, NCT00952029 . Registered 8 August 2009.

Long-Term Relative Survival after Stroke: The Dijon Stroke Registry.

OBJECTIVE: The aim of this study was to assess long-term survival after stroke and to compare survival profiles of patients according to stroke subtypes, age, and sex, using relative survival (RS) method. METHODS: All patients with a first-ever stroke were prospectively recorded in the population-based Dijon Stroke Registry from 1987 to 2016. RS is the survival that would be observed if stroke was the only cause of death. Ten-year RS was estimated using a flexible parametric model of the cumulative excess mortality rate, which was obtained by matching the observed all-cause mortality in the stroke cohort to the expected mortality in the general population. A separate model was fitted for each stroke subtypes, first fitted for each age and sex separately, and then adjusted for age and sex. RESULTS: In total, 5,259 patients (mean age 74.9 ± 14.3 years, 53% women) were recorded including 4,469 ischemic strokes (IS), 655 intracerebral hemorrhages (ICH), and 135 undetermined strokes. In IS patients, unadjusted RS was 82% at 1 year and decreased to 62% at 10 years. Adjusted RS showed a lower survival in older age groups (p < 0.001), but no difference between men and women (p = 0.119). In ICH patients, unadjusted RS was 56 and 42% at 1 and 10 years, respectively, with a lower adjusted survival in older age groups (p < 0.001), but no sex differences (p = 0.184). CONCLUSION: This study showed that RS after stroke is lower in older than in younger patients but without significant sex differences, and survival profiles differ according to stroke subtypes. Since RS allows a better estimation of stroke-related death than observed survival does, especially in old patients, such a method is adapted to provide reliable information when considering long-term outcome.

Outcomes following polypectomy for malignant colorectal polyps are similar to those following surgery in the general population.

OBJECTIVE: Population-based studies on colorectal malignant polyps (MPs) are scarce. The aim of this study was to describe time trends in the incidence of colorectal MPs before and after the introduction of a colorectal mass-screening programmein 2003 and to assess outcomes (survival and recurrence) after endoscopic or surgical resection in patients with MPs. DESIGN: We included 411 patients with MPs diagnosed between 1982 and 2011 in a well-defined population. Age-standardised incidence rates were calculated. Univariate and multivariate 5-year recurrence and net survival analyses were performed according to gross morphology. RESULTS: Age-standardised incidence of MPs in patients aged 50-74 years doubled from 5.4 in 1982-2002 to 10.9 per 100 000 in 2003-2011. Pedunculated MPs were more frequently resected endoscopically (38.2%) than were sessile MPs (19.1%; p<0.001). For patients with pedunculated MPs and a pathological margin ≥1 mm, the 5 -year cumulative recurrence rate did not differ significantly between surgical and endoscopic resection (8.2% and 2.4%, respectively). For patients with sessile MPs, it was 3.0% after first-line or second-line surgical resection, 8.6% after endoscopic resection and 17.9% after transanal resection (p=0.016). The recurrence rate decreased dramatically for patients with sessile MPs from 11.3% (1982-2002) to 1.2% (2003-2009) (p=0.010) and remained stable for pedunculated MPs at 4.6% and 6.7%, respectively. Five-year net survival was 81.0% when pathological margins were <1 mm and 95.6% when ≥1 mm (p=0.024). CONCLUSION: Outcomes following polypectomy in patients with a pathological margin ≥1 mm are similar to those following surgery in the general population. Endoscopic resection needs to be completed by surgery if pathological margins are less than 1 mm.

A retrospective analysis of the impact of treatments and blood counts on survival and the risk of vascular events during the course of polycythaemia vera.

Vascular and non-vascular complications are common in patients with polycythaemia vera. This retrospective study of 217 patients with polycythaemia vera aimed to determine whether blood counts with respect to different treatments influenced the complication rate and survival. We found that 78 (36%) patients suffered from at least one complication during follow-up. Older age and elevated lactate dehydrogenase at diagnosis were found to be risk factors for vascular complications. When the vascular complication occurred, 41% of the patients with a complication had elevated white blood cells (WBC) compared with 20% of patients without a complication (P = 0·042). Patients treated with hydroxycarbamide (HC; also termed hydroxyurea) experienced significantly fewer vascular complications (11%) than patients treated with phlebotomy only (27%) (P = 0·013). We also found a survival advantage for patients treated with HC, when adjusted for age, gender and time period of diagnosis (Hazard ratio for phlebotomy-treated patients compared to HC-treated patients at 5 years was 2·42, 95% confidence interval 1·03-5·72, P = 0·043). Concerning survival and vascular complications, HC-treated patients who needed at least one phlebotomy per year were not significantly different from HC-treated patients with a low phlebotomy requirement. We conclude that complementary phlebotomy in HC-treated patients in order to maintain the haematocrit, is safe.

High risk of cancer in autoimmune necrotizing myopathies: usefulness of myositis specific antibody.

Cancer can occur in patients with inflammatory myopathies. This association is mainly observed in dermatomyositis, and myositis-specific antibodies have allowed us to delineate patients at an increased risk. Malignancy is also reported in patients with necrotizing autoimmune myopathies, but the risk remains elusive. Anti-signal recognition particle or anti-HMGCR antibodies have been specifically associated with necrotizing autoimmune myopathies. We aimed at screening the incidence of cancer in necrotizing autoimmune myopathies. A group of patients (n = 115) with necrotizing autoimmune myopathies with or without myositis-specific antibodies was analysed. Malignancy occurred more frequently in seronegative necrotizing autoimmune myopathies patients and in HMGCR-positive patients compared to anti-signal recognition particle positive patients. Synchronous malignancy was diagnosed in 21.4% and 11.5% of cases, respectively, and incidence of cancer was higher compared to the general population in both groups. No specific type of cancer was predominant. Patients suffering from a synchronous cancer had a decreased median survival time. Cancer screening is necessary in seronegative necrotizing autoimmune myopathies and in HMGCR-positive patients but not in anti-signal recognition particle-positive patients.

Time trends in colorectal cancer early postoperative mortality. A French 25-year population-based study.

BACKGROUND: Postoperative mortality after resection of colorectal cancer is an important issue. The aim of this study was to assess early postoperative mortality in a well-defined French population. METHODS: Data on 30- and 90-day postoperative mortality after resection for colorectal cancer were extracted from the digestive cancer registry of Burgundy. Time trends of postoperative mortality between 1989 and 2013 were described for the large population. Case-control studies (death within 30 or 90 days = cases, alive at 90 days = controls) focused on the association between postoperative mortality and surgical approach, obesity and other comorbidities over the last [2010-2013] period, using conditional logistic regressions. RESULTS: Among the 11,448 concerned patients, 30- and 90-day postoperative mortalities were 4.9 and 7.2%. Thirty-day operative mortality decreased from 7.2% (1989-1993) to 4.4% (2010-2013; p < 0.001) for colon cancer and from 4.2 to 3.3% for rectal cancer (NS). Diagnosis before 1997, male gender, advanced age, emergency surgery and palliative resection were associated with a significantly higher 30- and 90-day mortality rate. The univariate risk of mortality was two to three times higher for conventional open laparotomy and conversion than for laparoscopy-assisted surgery. The surgical approach was no longer significant in multivariate analysis. Emergency surgery and comorbidities were associated with higher 30- and 90-day postoperative mortality, whereas obesity was not specific. CONCLUSION: Postoperative mortality after colorectal resection decreased over time. Surgical approach had no influence on early mortality. Improvement in the management of the elderly and patients with comorbidities is a challenge to reduce postoperative mortality in the future.

Pancreatic cancer: Wait times from presentation to treatment and survival in a population-based study.

Pancreatic survival is one of the worst in oncology. To what extent wait times affect outcomes in unknown No population-based study has previously explored patient and treatment delays among individuals with pancreatic cancer. The aim of this study was to estimate patient and treatment delays in patients with pancreatic cancer and to measure their association with survival in a nonselected population. All patients diagnosed with pancreatic cancer for the first time between 2009 and 2011 and registered in two French digestive cancer registries were included. Patient delay (time from onset of symptoms until the first consultation categorized into <1 or ≥1 month), and treatment delay (time between the first consultation and treatment categorized into less or more than 29 days, the median time) were collected. Overall delay was used to test associations between survival and the timeliness of care by combining patient delay and treatment delay. Patient delay was longer than 1 month in 46% of patients. A patient delay longer than one month was associated with the absence of jaundice (p < 0.001) and the presence of metastasis (p = 0.003). After adjusting for other covariates, such as symptoms and treatment, the presence of metastasis was negatively associated with treatment delay longer than 29 days (p = 0.025). After adjustment for other covariates, especially metastatic dissemination and the result of the resection, overall delay was not significantly associated with prognosis. We found little evidence to suggest that timely care was associated with the survival of patients.

Early Gastric Cancer: Trends in Incidence, Management, and Survival in a Well-Defined French Population.

BACKGROUND: Little is known about the epidemiology of early gastric cancer (EGC) in Western countries. The aim of this study was to analyze trends in the incidence, management, and survival of EGC in a well-defined population over a 30 year period. METHODS: Data were obtained from the population-based cancer Registry of Burgundy (France). Incidence rates were calculated by sex, age, and 10 year period of diagnosis. Net survival rates were calculated and a multivariate relative survival analysis performed. RESULTS: EGC represented 6.7 % of gastric cancer diagnosed between 1982 and 2011. Age-standardized incidence rates were higher in men (0.79/100,000) than in women (0.40/100,000). Between the periods 1982-1991 and 2002-2011, it decreased from 0.97 to 0.53 per 100,000 in men and from 0.44 to 0.30 per 100,000 in women. Overall, 19 % of the tumors were limited to the mucosa, 69 % to the submucosa, and 15 % invaded lymph nodes. Node invasion and male sex were the only significant prognostic factors. Five-year net survival was 50 % in node-positive patients and 85 % in node-negative patients (p < 0.001). In multivariate analysis, the relative risk of death in men compared to women was 2.3 and was 10.4 in patients with positive nodes compared to patients with negative nodes. CONCLUSIONS: EGCs are rare in France. The prognosis is favorable, except for node-positive cancers, which may benefit from the recently developed adjuvant chemotherapy for gastric cancer.

Incidence and patterns of late recurrences in colon cancer patients.

Long-term recurrences of colon cancer raised questions about the possible benefit of prolonging the recommended active 5-year surveillance. The aim of this study was to determine, for the first time, the incidence and patterns of late 10-year recurrence following curative resection of colon cancer. Data were obtained from two French digestive cancer registries. A total of 3,622 patients under 85 years resected for cure for colon cancer diagnosed between 1985 and 2000 were included. Information regarding recurrences was actively collected. Cumulative failure rates at 10 years were estimated using Kaplan-Meier estimates corrected by cause-specific hazards, and multivariable analysis was performed using a model for the subdistribution of a competing risk proposed by Fine and Gray. The overall cumulative recurrence rate between 5 and 10 years after initial surgery was 2.9% for local recurrence and 4.3% for distant metastasis. Among men with no recurrence 5 years after diagnosis of colon cancer, 1 in 12 developed a recurrence between 5 and 10 years, and the corresponding cumulative rate was 7.8%. The frequency was 1 in 19 for women, corresponding to a cumulative rate of 5.2%. In the multivariate analysis, non-emergency diagnostic feature, female sex and age under 75 were associated with a lower risk of recurrence. Stage at diagnosis was not a predictor of late recurrence. Late recurrence after colon cancer resection with curative intent can occur. A regular clinical follow-up is necessary to detect early signs of possible recurrence.

Incidence and patterns of late recurrences in rectal cancer patients.

BACKGROUND: Long-term recurrences of rectal cancer raised questions about the possible benefit of prolonging the recommended active 5-year clinical and endoscopic surveillance. The aim of this study was to determine for the first time, incidence and patterns of late 10-year recurrence after curative resection of rectal cancer. METHODS: The study included 1,222 patients with rectal cancer resected for cure between 1985 and 2000 from those registered in two French population-based digestive cancer registries. Information about local recurrences and distant metastases at 10 years was retrospectively and actively collected up to January 1, 2011. RESULTS: Although the overall 5-year cumulated rate was 39.5 %, the 10-year cumulated rate was 44.1 % (25.6 % for local recurrence and 29.9 % for distant metastases). In multivariate analyses, TNM stage was associated with a higher risk of local recurrence (hazard ratio [HR] stage III vs. stage I = 3.98 [95 % confidence interval, 2.66-5.94]) and of distant metastasis (HR = 3.60 [2.65-4.91]). Preoperative radiotherapy decreased the risk of local recurrence (HR = 0.43 [0.28-0.66]), but not the risk of metastasis. Patients diagnosed between 1995 and 2000 were less prone to develop long-term metastasis than those diagnosed between 1985 and 1989 (HR = 0.66 [0.49-0.88]). Among patients without recurrence 5 years after diagnosis, one patient in 13 developed a recurrence between 5 and 10 years. CONCLUSIONS: Late recurrences do exist. A personalised surveillance could be extended until 10 years according to the characteristics of primary tumour and the patient.

Trends in the management of gastric cancer over a 32-year period: a French population-based study.

BACKGROUND: Gastric cancer is still generating interest because of its poor prognosis. The aim of this study was to investigate time trends in diagnostic assessment, patterns of care, and survival of gastric cancers. METHODS: We considered 5,010 gastric cancers diagnosed between 1976 and 2007 in a well-defined French population. Logistic regressions were used to identify factors associated with R0 resection and operative mortality. A multivariate relative survival analysis was performed. RESULTS: Diagnostic modalities have changed. Since 1988, endoscopy is performed when gastric cancer is suspected (95.5%). However, there has been no strong variation in stage over time: the proportion of stage I cancers increased from 5.5% to 13.4% between the periods 1976-1979 and 2004-2007 (p < 0.001) whereas that of advanced cases remained stable, 64.8% and 65.0%, respectively. R0 resections rose from 36.7% (1976-1979) to 46.7% between 1980 and 1999, and decreased to 32.7% thereafter. Age, tumor location, and period were associated with R0 resection. Neoadjuvant and adjuvant chemotherapy were rarely used before 2000, then reached 15.0% and 19.1%, respectively, during the later period. Operative mortality after R0 resection decreased from 18.3% during the 1976-1979 period to 4.3% during the 2004-2007 period (p < 0.001). Prognosis slightly improved during the three first periods, from 13.0% to 22.6%, then leveled off, not exceeding 26.0% thereafter. Stage, age, histology, and time period significantly influenced survival. CONCLUSION: Changes in diagnostic modalities were associated with minor changes in stage and prognosis for gastric cancer. Earlier diagnosis and new therapeutic strategies are the best way to improve the prognosis.

Incidence and survival in late liver metastases of colorectal cancer.

BACKGROUND AND AIMS: Data concerning the risk of long-term liver metastasis following surgery of colorectal cancer in the general population are scarce. The 10-year incidence and prognosis of metachronous liver metastases remain unknown. METHODS: Among 4584 patients resected for cure for colorectal cancer recorded in two French digestive population-based cancer registries between 1985 and 2000, 602 presented metastases including liver metastases. RESULTS: The cumulated incidence of liver metastasis was 15% at 5 years and 17% at 10 years, and was mainly related to stage at diagnosis. The 10-year cumulative incidence was 6% for stage I and 30% for stage III. The hazard ratio was 3.2 [2.4-4.3] for stage II and 6.9 [5.1-9.2] for stage III compared with stage I. Among survivors with no recurrence five years after diagnosis, 2.2% developed liver metastasis between 5 and 10 years. Resection for cure of liver metastases was performed in 35% of patients aged under 75 years and in 10% of patients over 75 (P < 0.001). After resection for cure, 10-year relative survival improved from 21% during the period 1985-1997 to 34% during the period 1998-2011 (P = 0.023). Survival in patients with liver metastasis diagnosed between six and 12 months after surgery was less than half that in patients with metastasis diagnosed later (HR: 0.6 [0.4-1.0]). CONCLUSION: Liver metastases from colorectal cancer remain a substantial problem and continue to occur long after five years. This study furnishes unbiased figures that can be used as a reference. Liver metastases that appear late have a better prognosis.