Abstract
The aim of this study was to evaluate predictors of survival in stage IV metastatic colorectal cancer (CRC). Patients and Methods: A total of 541 patients with histologically proven metastatic CRC (UICC stage IV) were retrospectively analysed and 37 variables were tested for their potential relationship to survival. Results: Mean survival time was recorded at 12.8 months [95% confidence interval (CI) 12.0-13.5]. Three factors were independently associated with improved survival: combination chemotherapy, improved performance status and dermatological complications. Eight factors were independently associated with unfavorable survival: worsened performance status, C-reactive protein >5 mg/dl, anemia, anorexia, weight loss ≥10%, fatigue, hypoalbuminemia and blood transfusions. Conclusion: A number of factors could be used as predictors of survival in patients with stage IV metastatic CRC. Patients who are relatively fit, have low CRP levels and tolerate combination chemotherapy appear to have a more favorable survival outcome.
Approximately 35% of colorectal cancer (CRC) patients present with stage IV metastatic disease at the time of diagnosis and 20%-50% with stage II or III disease will progress to stage IV at some point during the course of their disease (1-3). Stage IV CRC carries a dismal prognosis: the 5-year survival rate for stage IV CRC is less than 10% (2, 3) and the median survival time of patients with stage IV CRC given optimal supportive care without chemotherapy is approximately 5 months (4).
In metastatic CRC, chemotherapy is used mainly with palliative intent. It improves quality of life and prolongs survival in comparison with best supportive care alone (4). Until a few years ago, 5-fluorouracil (5-FU) modulated with folinic acid was the reference first-line treatment option for metastatic CRC, with objective response rates of 10-25% and managable toxicity (5). Treatment options for patients with metastatic CRC are continuously evolving, with the development of new chemotherapeutic agents. With three classes of cytotoxic agents and two classes of therapeutic antibodies currently available, treatment decision-making is more complicated as the optimum sequencing and dosing of cytotoxic and biological agents remains to be determined. Nevertheless, during the past decade, the application of these new chemotherapeutic agents in the treatment of metastatic CRC has increased, with concomitant improvement in outcome (6, 7). However, as treatment with these new agents is associated with significant toxicity, there is a need to identify prognostic factors which may determine treatment response and survival. Such an approach could refine management with palliative chemotherapy according to the likelihood of clinical benefit (8). The aim of this study was to evaluate predictors of survival in stage IV metastatic CRC.
Patients and Methods
Patients. The medical records of 541 patients, with histologically proven metastatic CRC (UICC stage IV) between 1998 and 2008 were retrospectively reviewed. All were consecutive non-selected cases from a single center and all patients were treated outside of clinical trials. No patients were candidates for surgical treatment (either curative or palliative), however, all received palliative chemotherapy. Records with complete date (for the parameters used as prognostic factors) were included in the analysis. Single-agent chemotherapy regimens were based on leucovorin modulated 5-FU (Mayo clinic or AIO regimens). Combination chemotherapy regimens included combination treatments of 5-FU (DeGrammont or simple infusion and leucovorin) with either oxaliplatin or irinotecan, or capecitabine with or without bevacizumab or cetuximab. Patients were monitored periodically for clinical and laboratory evidence of toxicity.
Prognostic variables. A total of 37 patient-related factors, tumor-related factors and complications related to either disease progression or treatment were entered in the analysis based on factors identified by previous studies (9, 10), as well as our own clinical experience. Patient-related factors included age (≤60 years or >60 years), gender and performance status (PS) according to the Karnofsky Performance Status Scale Index. Tumor-related factors included symptoms: fever (yes vs. no), pain (yes vs. no); location of metastases: lymph nodes (yes vs. no), liver (yes vs. no), lung (yes vs. no), abdomen/peritoneum (yes vs. no), pelvis (yes vs. no), local recurrence (yes vs. no), bone (yes vs. no), skin (yes vs. no), adrenal glands (yes vs. no); and biochemical parameters. For the latter, group categorizations were used: for carcinoembryonic antigen (CEA): normal ≤5 mg/dl and elevated >5 mg/dl; for cancer antigen 19-9 (CA 19-9): values ≤30 × normal (NL) vs. >30 × NL; for C reactive protein (CRP): normal <5 mg/dl, moderately elevated 5-15 mg/dl, highly elevated >15 mg/dl. Factors related to either response to treatment or disease progression included chemotherapy (single agent vs. combination chemotherapy), change in PS (no change, improved, worse) and hematological complications. For the latter, group categorizations were used: for neutropenia: white blood cell count >4000/μl, 2000-4000/μl, <2000/μl; for anemia: hemoglobin <8.5 g%, 8.5-10 g%, 10-12 g%, 12-13.5 g%, >13.5 g%; for thrombocytopenia: yes, platelets (PLT) ≤100,000/μl vs. no, PLT>100,000/μl. Other factors included nausea-vomiting (yes vs. no), diarrhea (absent, frequent, continuous problem), anorexia (absent, frequent, continuous problem), weight loss ≥10% (yes vs. no), mucositis (absent, frequent, continuous problem), fatigue (absent, frequent, continuous problem), neurological complications (yes vs. no), psychological complications (yes vs. no), alopecia (yes vs. no), hand and foot syndrome (yes vs. no), dermatological complications (yes vs. no), liver toxicity (yes vs. no), use of epoetin (yes vs. no), blood transfusions (yes vs. no) and hypoalbuminemia (yes vs. no).
Statistical analysis. Descriptive statistics were calculated with the use of mean, median and standard deviation for quantitative measurements and counts/percentages for discrete factors. Overall survival was defined as the time from the first day of diagnosis of metastatic CRC disease to death of any cause, or to the last follow-up examination. Survival data were studied with the use of Kaplan-Meier method. Changes in survival between groups were recorded with the log-rank test. Additionally, Wald statistics were implemented for bivariate associations between survival and quantitative measures such as age. A multivariate Cox regression model was implemented for the study of the parallel effect of remaining parameters on survival. Best model selection was based on automated techniques. Regression results were displayed in the form of regression estimates tables. Hazard ratios of outcomes under study were calculated for each parameter estimate as well as 95% confidence intervals. Categorical covariates were compared with a predefined reference category group. All analyses were conducted at a predefined significance level of 5% with the use of statistical package SPSS 12.0.
Results
Patients. A total of 541 patients were included in the study, with median age of 61.00 years (range 38 years) a mean age of 60.33 years and standard deviation of 7.35 years. The frequencies of the clinical variables are shown in Table I.
Overall survival. Survival data were collected for all patients. Based on the Kaplan-Meier method, the mean survival time was recorded at 12.8 months [95% confidence interval (CI) 12.0-13.5 months], with a median survival of 9.8 months (95% CI 8.8-10.8 months) (Figure 1).
Bivariate analysis. Improved PS, combination chemotherapy, elevated CRP, fever, pain, abdominal/peritoneal metastasis, local recurrence, anemia, anorexia, fatigue, weight loss ≥10%, thrombocytopenia, hypoalbuminemia, skin complications and blood transfusions were significantly associated with worse survival (p<0.001 for all, with the exception of abdominal/peritoneal metastasis, where p=0.02).
Multivariate analysis. Factors found to have strongest significance of a relation to survival according to the bivariate analysis were entered into the multivariate analysis model. Factors were added and excluded using the change in likelihood between models as inclusion and exclusion criteria. Forward automated procedures resulted in the final model, which is described in Table II.
Hazard ratios of risk factors. Patients with improved PS had 1.4 times lower risk of death than patients with no change of PS, while those with worsened PS had 2.6-fold higher risk of death than patients with no change in PS (Figure 2A). Patients who received combination chemotherapy had 2.6-fold lower risk of death than those who did not (Figure 2B). Probability of death increased as CRP increased; patients with CRP >15 mg/dl had 1.48-fold higher risk of death and patients with CRP 5-15 mg/dl had 1.37-fold higher risk of death than patients with CRP <5 mg/dl (Figure 2C). Anemia was also associated with higher risk of death. In particular, patients with Hb 10-12 g%, Hb 8.5-10 g% and Hb <8.5 g% had 1.58-, 1.33- and 1.38-fold respectively higher risk of death than patients with Hb >13.5 g% (Figure 2D). Patients with weight loss ≥10% had 3.3-fold higher probability of death than patients without (Figure 3A). The risk of death increased as anorexia became a more severe symptom. Patients who described anorexia as frequent problem had a 1.21-fold higher risk of death than patients without this symptom, while those patients who described anorexia as continuous problem had 2.64 times higher risk of death (Figure 3B). Similarly, the risk of death increased as fatigue became more prominent. Patients with fatigue as a frequent problem had 1.75-fold higher risk of death and patients with fatigue as a continuous problem had 2.56-fold higher risk of death than patients not reporting this symptom (Figure 3C). Patients with dermatological complications had 1.26 times lower probability of death than did those without (Figure 3D). Patients with hypoalbuminemia had 1.27 times higher probability of death than did those without (Figure 4A) and finally the need for blood transfusions was associated with a 2-fold higher probability of death (Figure 4B).
Demographic, clinical and laboratory variables in the study population (n=541).
Discussion
Over the last few years there has been an increased interest in clinical and molecular prognostic factors in stage IV metastatic CRC. One reason for this is the recent advances in treatment options which have extended patient survival. A second reason is the difference in survival for patients who receive chemotherapy in randomized trials. It is anticipated that the latter could be explained by differences in patient selection and study design, since although patient stratification in randomized trials is based on selected prognostic factors such as PS, the influence of some other prognostic factors or their combination could abrogate any potential survival benefit from the antineoplastic agent or drug combination tested (10).
Overall survival (n=541).
Our analysis included consecutive non-selected stage IV metastatic cololorectal cancer cases from a single center and all patients were treated outside of clinical trials with palliative chemotherapy only. Based on the above setting, we distinguished three groups of factors that influence survival. The first group of factors relates to clinical and laboratory parameters depicting patient functional status: worsened PS, anorexia, fatigue, weight loss, hypoalbuminemia, anemia and need for blood transfusions, which all had a negative influence on survival. On the other hand dermatological complications had a positive influence on survival. The second group of factors relates to disease burden and subsequent inflammatory response: increased CRP had negative influence on survival. The third group of factors relates to therapy: combination chemotherapy had positive influence on survival.
Various studies on survival factors for stage IV metastatic CRC have resulted in disparate results which probably depict differences in patient population and study design and hence make them difficult to evaluate. Nevertheless, in these studies PS is established as a fundamental clinical prognostic factor in metastatic CRC (11-23). Worsening PS has been definitively associated with advanced disease and poor prognosis and this is confirmed in our analysis of an unselective pool of patients with stage IV CRC. Anorexia and fatigue are commonly associated with poor general status in cancer patients and may be related to the disease itself, drug toxicity or both. Their predictive value in CRC has not been investigated before; hence, our study is the first to validate their prognostic significance in stage IV metastatic CRC. Weight loss is yet another factor associated with poor functional capability and has been shown before to be independently associated with survival in patients with advanced CRC included in irinotecan phase III trials (20). The association between anemia and inferior survival capability has also been demonstrated by Köhne et al. previously in their multivariate analysis of 3,825 patients with stage IV CRC treated with palliative 5FU-based chemotherapy in the setting of 22 multinational trials (18) and the use of epoetin has been justified to reduce the need for blood transfusions (24, 25). Serum albumin reflects the nutritional status of patients, furthermore it also depicts general condition, including reserve capacity and has also been shown before to have a prognostic significance for survival in patients with metastatic CRC (12, 14, 16, 22, 26).
Final Cox proportional odds regression model.
Survival data according to performance status (A), combination chemotherapy (B), CRP (C) and anemia (D).
Skin toxicity was previously correlated to survival. In a phase II study including 346 CRC patients treated with cetuximab who were refractory to irinotecan, oxaliplatin, and fluoropyrimidines, a partial response occurred in 7%, 17%, and 20% of patients with grade 1, 2, and 3 skin rash, respectively. Survival also related strongly to the severity of the rash (27).
The role of CRP as a predictor of survival in CRC is controversial. CRP is a nonspecific but sensitive marker of inflammation. Interleukin-6 (IL-6), IL-8, and tumor necrosis factor alpha induce the synthesis of CRP in hepatocytes. CRP has been positively correlated with weight loss, anorexia-cachexia syndrome, extent of disease, and recurrence in advanced cancer including colorectal cancer (28, 29). However Wigmore et al. reported that there was no difference in Dukes' stage between patients with or without an elevated level of CRP (30) and Chung et al. performed a multivariate analysis of 172 patients with CRC and concluded that CRP level is not an independent prognostic factor for survival (31). In these studies, the prognostic significance of CRP was examined in a mixed population including patients at various stages of CRC, hence our study is the first to validate the prognostic importance of CRP for stage IV metastatic CRC.
Chemotherapy is a well established palliative treatment strategy in stage IV metastatic CRC and has been shown to independently predict survival (15, 21, 26, 32). Combination chemotherapy regimens including irinotecan and oxaliplatin in combination with 5-FU, with or without a biological agent, have improved response rates to as high as 50% and overall survival times to 15-20 months. Irinotecan-based regimens were shown to be independently associated with a better survival in patients with advanced or colorectal cancer (20, 33). Our analysis confirms the importance of combination chemotherapy as an independent predictor of survival for metastatic stage IV metastatic CRC.
Survival data according to weight loss ≥10% (A), anorexia (B), fatigue (C) and dermatological complications (D).
Survival data according to hypoalbuminemia (A) and transfusion (B).
In conclusion, with the present study, we define predictors of survival for patients with stage IV metastatic CRC. Patients who are relatively fit, have low CRP levels and tolerate combination chemotherapy appear to have a more favourable survival outcome. This information should be considered for the selection of appropriate management strategies for patients with stage IV metastatic CRC and the design of future clinical trials.
Acknowledgements
We thank Mrs Madalena Zlovotska for data collection and Mr Dimitrios Boulamatsis for performing the statistical tests.
Footnotes
- Received August 2, 2009.
- Revision received January 20, 2010.
- Accepted January 21, 2010.
- Copyright© 2010 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- Elucidating Tumor-stromal Metabolic Crosstalk in Colorectal Cancer through Integration of Constraint-Based Models and LC-MS Metabolomics
- Serine Threonine Kinase 17A Maintains the Epithelial State in Colorectal Cancer Cells
- A Routine Laboratory Data-Based Model for Predicting Recurrence After Curative Resection of Stage II Colorectal Cancer
- Significance of Markers of Systemic Inflammation for Predicting Survival and Chemotherapeutic Outcomes and Monitoring Tumor Progression in Patients with Unresectable Metastatic Colorectal Cancer
- Reply to F.V. Raimundo et al