Time trends in the treatment and survival of recurrences from colorectal cancer

F. Guyot, J. Faivre, S. Manfredi, B. Meny, C. Bonithon-Kopp and A. M. Bouvier*

Registre Bourguignon des Cancers Digestifs (INSERM EPI 0106 and CIC-EC01), Faculté de Médecine, Dijon Cedex, France

* Correspondence to: Dr A.-M. Bouvier, Registre Bourguignon des Cancers Digestifs (INSERM EPI 0106), Faculté de Médecine, BP 87 900 21079 Dijon Cedex, France. Tel: +33-3-80-39-33-40; Fax: +33-3-80-66-82-51; Email: anne-marie.bouvier{at}u-bourgogne.fr


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Background:: Little is known about the management of recurrences from colorectal cancer at a population level.

Materials and methods:: Data was obtained from the population-based cancer registry of Côte d'Or (Burgundy, France) over a 28-year period. Univariate and multivariate analyses were performed to analyse trends in treatment and survival for local recurrence and distant metastases.

Results:: The proportion of patients resected for cure increased from 6.7% (1976–1984) to 23.7% (1994–2003; P <0.001) for distant metastases and from 15.9% to 58.1% (P <0.001) for local recurrence. Age and period of diagnosis were independent factors associated with a resection for cure. Rectal cancer local recurrence was less often resected for cure than colon cancer local recurrence (P=0.05). Long-term survival was observed only after resection for cure: 5-year relative survival rates were 36.1% for local recurrence and 24.0% for distant metastases. In the multivariate analysis, survival decreased with age and increased over time but significantly only over the last study period. Surgical resection and palliative chemotherapy were other determinants of prognosis for distant metastases whereas surgical resection and palliative radiotherapy did influence the prognosis for local recurrence.

Conclusion:: Substantial advances in the management of recurrences have been achieved over time. More effective treatments and mass screening represent promising approaches to decrease this problem.

Key words: colorectal cancer, distant metastases, local recurrence, cancer registry, survival, treatment


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
It is estimated that 37 000 new cases of colorectal cancer will occur each year in France, representing 15% of all cancers [1Go] and that approximately 85% will be resected for cure [2Go]. It is estimated that about one-third will develop a metachronous recurrence during the following 5 years [3Go–5Go]. This severely impairs quality of life and is associated with a high risk of mortality. Over the past few years, there have been many changes in the management of colorectal recurrences [6Go]. These include diagnostic procedures, extension of the indications of resection for recurrences, improvement in peri-operative care and development of effective palliative treatments. However, the impact of these improvements on the overall population is not known. There is a paucity of reports on their management and prognosis over time. Most available reports are hospital-based. Because of selection bias, especially when considering surgical series from specialised centres, these data cannot be used as reference values. The outcomes of the management of recurrences from colorectal cancer at a community level are not known. Such studies are difficult to perform because they require the active participation of the entire medical profession and accurate data, which is seldom collected by cancer registries. The aim of this study was to draw a picture of time trends in treatment and survival of both local recurrence and distant metastases from colorectal cancer in a well-defined French population.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Population
All digestive tract cancers in the Côte-d'Or, Burgundy are recorded in a population-based cancer registry (507 000 inhabitants according to the 1998 census). Cancer registration began in 1976. Information is regularly obtained from pathologists, hospitals (university hospitals including the comprehensive cancer centre, general hospitals), private physicians (gastroenterologists, surgeons, oncologists and radiotherapists) and general practitioners, as well as from the National Health system and a monthly review of death certificates. Because of the multiplicity of information sources, it was assumed that nearly all newly diagnosed cancers had been registered. The quality and comprehensiveness of registration is certified every 4 years by an audit of the National Institute of Health and Medical Research (INSERM) and of the National Public Health Institute (InVS).

The data routinely collected is related to diagnostic strategies, treatment, stage at diagnosis and follow-up of the patients. Data on metachronous metastases and local recurrences is not collected routinely. Three surveys were conducted. The first one was performed in 1987 and dealt with patients diagnosed between 1976 and 1984, the second one in 2000 with patients diagnosed between 1985 and 1995, and the last one in 2003 with patients diagnosed between 1996 and 2000. Information about recurrence was obtained from all clinicians (general practitioners and specialists) involved in the management and the follow-up of these patients.

Studied variables
We reviewed a series of 3655 patients resected for cure for colorectal cancer in the Côte d'Or area between 1976 and 2000. Information on recurrence was available for 3492 patients (95.5%) in January 2004. The follow-up period was between 3 and 28 years. Overall, 1064 patients presented a recurrence. For all patients, the studied variables included: sex, age, date of diagnosis of the recurrence, site of recurrence (local recurrence, distant metastases) and its treatment. Date at diagnosis of recurrence was tabulated in three periods: 1976–1984, 1985–1993 and 1994–2003. Age at recurrence was classified into two categories: <75 years and ≥75 years. Only recurrences diagnosed after the first 3 months following diagnosis of the primary cancer were included in the present study. Two groups were distinguished: local recurrence (regrowth of tumour in and around the tumour bed) and distant metastases. Patients with both local recurrence and distant metastases (i.e. the two events diagnosed within 30 days) were considered as multiple organ involvement in the ‘metastases’ group (n=92). The exact location of distant metastases remained unknown for 37 cases. Location of the primary tumour was divided, according to the International Classification of Diseases for Oncology, 3rd revision [7Go], into colon (caecum to rectosigmoid junction; C-18 and C-19) and rectum ampulla (C-20). Treatment was divided into five categories: surgery for cure (removal of all macroscopic detectable disease and clear resection margins), palliative resection (failure to resect all disease), palliative chemotherapy, radiotherapy without resection (for rectal cancer) and symptomatic treatment when there was no carcinolytic intention including derivation and laparotomy and antalgic radiotherapy. For 70 cases, treatment procedures remained unknown (nine cases with local recurrences, 61 cases with distant metastases).

Statistical method
Associations between categorical data and local recurrence and distant metastases were performed using the {chi}2 test. A non-conditional logistic regression was used to identify risk factors independently and significantly associated with the probability of surgery for cure for local recurrence and for distant metastases. The categories of variables were represented by indicator variables and their predictive value was assessed with the P value of the log likelihood. After adjustment for age and sex, only variables that significantly improved the fit of the model (P <0.10) were included in the final model.

The life status was known for 1021 patients (96%) in January 2004. Crude and relative 5-year survival rates were also calculated. These were defined as the ratio of the observed survival rate to the expected survival rate in a population with similar sex and age distribution derived from local mortality tables. A multivariate analysis was performed using a relative survival model with proportional hazard applied to the net mortality by interval. This model makes it possible to calculate relative risks of death in comparison to a baseline which is the cumulative net hazard calculated from a priori defined interval [8Go]. Relative survival was computed using the RELSRUV program [9Go]. The significance of the covariates was tested by the likelihood ratio.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Treatment
A total of 338 patients with local recurrences and 726 patients with distant metastases were considered. The proportion of patients with distant metastases who underwent a resection for cure is given in Table 1. It increased between the first study period (1976–84) and the second (1985–93) and then remained stable (P <0.001). Overall, 26.4% of patients aged <75 years were resected for cure versus 8.4% of patients aged ≥75 years (P <0.001) (not shown in the table). Among patients resected for cure, 8.3% received adjuvant chemotherapy during the first study period, 8.6% during the second and 38.7% during the third. The proportion of patients receiving palliative chemotherapy was 8.0%, 8.6% and 30.2%, respectively (P <0.001). Palliative chemotherapy was carried out more often in patients with liver metastases alone (22.8%) or other metastatic site alone (28.1%) than in patients with multiple organ involvement (17.8%) (P=0.009). The proportion of symptomatically treated patients decreased dramatically over time, from 74.9% during the first study period to 37.5% during the last study period (P <0.001).


View this table:
[in this window]
[in a new window]
 
Table 1. Time trends in treatment (%) of distant metastases and local recurrence for colorectal cancer

 
The proportion of patients resected for cure varied with the site of the distant metastases. This was 45.4% for nodes metastases (5/11), 33.3% for ovary metastases (1/3), 24.5% for liver metastases (61/249), 24.0% for peritoneal dissemination whatever the extension (24/100), 18.3% for lung metastases (11/61), 5.9% for brain metastases (1/17) and 0% for bone metastases (0/16). Overall, 3.6% of patients with multiple organ involvement or with both distant metastases and local recurrence were resected for cure (30/221). Among patients with a metastasis limited to the liver, the proportion of patients resected for cure over the three study periods was 6.4%, 30.4% and 30.7%.

Concerning patients with local recurrence (Table 1), the resection-for-cure rate dramatically increased after the first period, reaching 58.1% during the last period. Thirty-four per cent of patients aged ≥75 years presenting a local recurrence were resected for cure versus 52.9% for younger patients (P=0.001). The proportion of patients resected for cure for a local recurrence was higher in the case of primary colon cancer (49.8%) than in the case of primary rectal cancer (38.3%; P=0.04). Palliative chemotherapy remained rarely used over the study periods. Overall, 14.8% of patients with rectal cancer and local recurrence underwent palliative radiotherapy. The proportion of patients with rectal cancer treated with radiotherapy alone strongly decreased over time, being 31.2%, 8.7% and 5.9%, respectively.

A multivariate logistic model was then applied to distant metastases and to local recurrence to identify factors independently related to surgery for cure. Patients <75 years were more likely than older patients to undergo a resection for cure. The period of diagnosis of the recurrence was also a factor strongly associated with a resection procedure. The primary location of the cancer played a role only for local recurrences. Rectal cancer recurrences were less often resected for cure than colon cancer recurrences (Table 2).


View this table:
[in this window]
[in a new window]
 
Table 2. Factors associated with resection for cure of recurrences from colorectal cancer (logistic regression)

 
Survival
The overall 1- and 5-year relative survival rates for distant metastases were 34.5% and 5.5%, and for local recurrence were 50.2% and 15.7%. In this last group, survival rates were, respectively, 52.4% and 16.2% for colon cancer local recurrence and 46.6% and 15.6% for rectal cancer local recurrence (NS). One- and 5-year survival rates according to sex, age, treatment and period of diagnosis of recurrence for distant metastases and local recurrences are given in Table 3. Survival rates were lower in patients with distant metastases than in patients with local recurrence only. There was no sex difference in prognosis. Survival rates were higher in patients <75 years than in older patients and improved over time for both distant metastases and local recurrence (P <0.0001). The increase in survival was more important for local recurrence than for distant metastases. Resection for cure was the most important determinant of survival (Figure 1). Long-term survival was seen only after resection for cure: 5-year survival rates were 21.6% for distant metastases and 31.6% for local recurrence. Compared to symptomatically treated patients, the survival curve of patients receiving chemotherapy was better for the first 3 years following diagnosis.


View this table:
[in this window]
[in a new window]
 
Table 3. Relative survival rates (%) for distant metastases and local recurrence by sex, age, period and treatment from colorectal cancer

 


View larger version (14K):
[in this window]
[in a new window]
 
Figure 1. Relative survival of patients with distant metastases from colorectal cancer according to treatment.

 
In the multivariate analysis, both for distant metastases and local recurrence, age, period at diagnosis of recurrence and treatment were independent significant factors (Table 4). Survival decreased with age and improved over time, but only significantly over the last study period. Surgical resection and palliative chemotherapy were other determinants of prognosis for distant metastases, whereas surgical resection and palliative radiotherapy did influence the prognosis for local recurrence. The relative risk of death was two to four times higher in the case of palliative cancer treatment than after surgery for cure.


View this table:
[in this window]
[in a new window]
 
Table 4. Factors of prognostic significance in distant metastases and local recurrence from colorectal cancer (multivariate relative survival model)

 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Data from this study have the advantage of including all cases diagnosed in a well-defined French population during a long time period, including those not operated on. Its aim was to provide non-biased and detailed statistics on the management of colorectal recurrences over a 28-year period. Information on treatment and survival was available for 96% of recorded cases. This study was carried out, therefore, without the referral bias that often occurs in hospital-based series. Many studies concerning the risk of recurrences of colorectal cancer have been published, but there is a paucity of reports concerning the medical care of metachronous events. All were hospital-based studies and often limited to surgically treated patients. This is the first report dealing with trends over time in the management of local and distant recurrences of colorectal cancer in a general population. One of the difficulties of assessing the outcomes of treatment for colorectal cancer in metachronous recurrences is that there is no consistent definition of the metachronous events in published literature. We considered as recurrences those diagnosed more than 3 months after diagnosis.

One of the main results of this study was the increase in the curative resection rate for distant metastases and above all for local recurrences when comparing the first study period with the two following periods. There was not much change over the 1985–2003 period. This increase remained significant after adjustment for sex, age at diagnosis and site of recurrence. The most common site of recurrence was the liver. The liver metastasis resection rate remained stable over the two last study periods, around 30% of cases. Rare published series dealt with hospital-based data and often analysed synchronous and metachronous liver metastases together. Another frequent localisation is peritoneal seeding. In our series, 24% of peritoneal dissemination was resected for cure. Our data did not allow an isolated nodule to be distinguished from peritoneal carcinomatosis, which explains the relatively high curative resection rate for this localisation. From this study, lung metastases were the third most frequent non-multiple organ involvement of distant metastases. As in other series, a few patients did undergo a curative resection of their lung metastases [10Go]. Other distant metastasis localisations were rare and the variability of the frequency of surgical resection was higher (nodes, ovary) or lower (brain, bone) than for liver metastases. The resection-for-cure rate was barely higher for colon cancer local recurrence than for rectal cancer local recurrence. Surgical curative techniques for colon cancer regrowth are easier than those for rectal cancer. Unsurprisingly, the proportion of patients with recurrences resected for cure was higher before the age of 75 than for older patients.

The most striking trend lies in the major improvement in survival over the study period. This finding is consistent with the increase in the proportion of patients resected for cure. Surgical resection was the only treatment that was associated with a substantial chance to survive 5 years. The overall 5-year survival rate was 31.6% for local recurrence and 21.6% for distant metastases. In the case of liver metastases resected for cure, it was 29.7%. This survival rate is comparable with that published in major collected series, which ranged between 21% and 33% [11Go–13Go]. However, 5-year survival rates were higher for the third study period compared with the second, although the resection-for-cure rate was similar. The improvement in prognosis remained significant after adjustment for age, sex, site of primary cancer and treatment. It was not related to a trend towards more favourable site of metastasis. There was no change over time in the proportion of the different sites of distant metastases, and adjustment on the site of metastases did not modify the prognosis effect of the period of diagnosis. The improvement of prognosis in this study period in patients with distant or local recurrence can be due to a more favourable stage at diagnosis of recurrences. This shows that major improvement in the management of patients with colorectal cancer recurrences has been achieved.

Prognosis of patients with unresected metastasis remains poor. However, trials published in the early 1990s demonstrated that chemotherapy was able to prolong survival and improve quality of life over best supportive care [14Go, 15Go]. This benefit of chemotherapy is observed over the first 3 years, when survival rates do not differ from symptomatically treated patients. Taking into account available results, the French consensus conference on colon cancer in 1998 recommended, when resection was not possible, the use of chemotherapy [16Go]. The publication of these data has led to a change in practice. There was an increase in the proportion of patients with distant metastases receiving chemotherapy over the 1994–2003 period compared with the previous periods. However, chemotherapy has not yet reached its full development. Only 30.2% of non-resected patients with distant metastases received this treatment over the 1994–2003 period. Our study indicates that there is a delay between the publication of scientific evidence and the complete implementation of effective treatment. Such information can only be provided by this type of population-based study.

Palliation with radiotherapy is possible for rectal cancer recurrence. However, long-term survival is rare. Radiotherapy for rectal cancer recurrences concerned 47.5% of non-resected cases. However, its practice decreased over time with the increase in the number of resections for cure performed.

Colorectal cancer recurrences remain a serious problem after resection for cure. However, resection of distant metastases or local recurrence provides a significant chance for cure to a subset of patients. Our results support a forceful surgical approach concerning recurrences whenever possible; in the short term, better surveillance schemes in order to detect recurrences as early as possible, the use of new and more effective chemotherapy regimens, refined immunotherapy protocols and approaches, requiring further investigation in prospective studies. Screening represents another promising approach to decrease the incidence of recurrences.


    Acknowledgements
 
This work was supported by a grant from the Foundation of France. We would like to thank Stéphanie Normand for statistical assistance.

Received for publication September 1, 2004. Revision received December 13, 2004. Accepted for publication December 20, 2004.


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
1. Bouvier AM, Remontet L, Jougla E et al. Incidence of digestive cancers in France. Gastroenterol Clin Biol 2004; 28: 877–881.[ISI][Medline]

2. Faivre-Finn C, Bouvier-Benhamiche AM, Phelip JM et al. Colon cancer in France: evidence for improvement in management and survival. Gut 2002; 51: 60–64.[Abstract/Free Full Text]

3. Devesa JM, Morales V, Enriquez JM et al. Colorectal cancer. The bases for a comprehensive follow-up. Dis Colon Rectum 1988; 31: 636–652.[ISI][Medline]

4. Abulafi AM, Williams NS. Local recurrence of colorectal cancer: the problem, mechanisms, management and adjuvant therapy. Br J Surg 1994; 81: 7–19.[ISI][Medline]

5. Manfredi S, Benhamiche AM, Meny B et al. Population-based study of factors influencing occurence and prognosis of local recurrence after surgery for rectal cancer. Br J Surg 2001; 88: 1221–1227.[CrossRef][ISI][Medline]

6. Scheele J, Altendorf-Hofmann A. Resection of colorectal liver metastases. Langenbecks Arch Surg 1999; 384: 313–327.[CrossRef][ISI][Medline]

7. International Classification of Diseases for Oncology, 3rd revision. Geneva: WHO 1995.

8. Estève J, Benhamou E, Croasadale M et al. Relative survival and the estimation of the net survival: elements for further discussion. Stat Med 1990; 9: 529–538.[ISI][Medline]

9. Hedelin G. Relsurv: a program for relative survival. Technical report of the Department of Epidemiology and Public Health, Faculty of Medicine. Strasbourg: Louis Pasteur Universtity 1999.

10. Penna C, Nordlinger B. Colorectal metastasis (liver and lung). Surg Clin North Am 2002; 82: 1075–1090.[CrossRef][ISI][Medline]

11. Nordlinger B, Benoist S. Recent advances in the case management of colorectal cancer liver metastases. Bull Acad Natl Med 2003; 187: 899–904.[ISI][Medline]

12. Hughes KS, Simon R, Songhorabodi S et al. Resection of the liver for colorectal carcinoma metastases: a multi-institutional study of patterns of recurrence. Surgery 1986; 100: 278–284.[ISI][Medline]

13. van Ooijen B, Wiggers T, Meijer S et al. Hepatic resections for colorectal metastases in The Netherlands. A multiinstitutional 10-year study. Cancer 1992; 70: 28–34.[ISI][Medline]

14. Nordic Gastrointestinal Tumor Adjuvant Therapy Group. Expectancy or primary chemotherapy in patients with advanced asymptomatic colorectal cancer: a randomized trial. J Clin Oncol 1992; 10: 904–911.[Abstract]

15. Scheithauer W, Rosen H, Kornek GV et al. Randomised comparison of combination chemotherapy plus supportive care with supportive care alone in patients with metastatic colorectal cancer. BMJ 1993; 306: 752–755.[ISI][Medline]

16. Conférence de Consensus. Prévention, dépistage et prise en charge des cancers du côlon 1998; 22: S275-S288.