5-Fluorouracil or capecitabine in the treatment of advanced colorectal cancer: a pooled-analysis of randomized trials
The aim of this pooled-analysis is to evaluate the benefit of capecitabine (C) versus standard intravenous 5-Fluorouracil (5-FU) as monochemotherapy or combination therapy in advanced colorectal cancer (CRC) in terms of safety and efficacy. Eligible patients have been randomized to receive either C-based or 5-FU-based chemotherapy for the treatment of advanced CRC. Relative risks (RRs) with 95% confidence intervals (CIs) of selected side effects (diarrhea, nausea, vomiting, stomatitis, hand and foot syndrome, neutropoenia, febrile neutropoenia, and cardio toxicity) and overall response rate (ORR) were calculated and hazard ratios (HRs) of progression-free survival (PFS) and overall survival were obtained, respectively, from published data. The RRs of stomatitis and neutropoenia are 0.39 and 0.40, respectively with C (P < 0.00001). In particular high-grade mucositis and neutropoenia, they are reduced by 69 and 74%, respectively (RR: 0.31 and 0.26). Diarrhea, nausea, vomiting, febrile neutropoenia, and cardio toxicity with C are not worse than with 5-FU. The RR of hand and foot syndrome with C compared to 5-FU is 3.45, (P < 0.00001). Response rate, PFS, and OS are equivalent in both C- and 5-FU-based regimens. The use of C instead of 5-FU in advanced colorectal cancer regimens results in significantly less toxicity in terms of stomatitis and neutroponenia. Only hand and foot syndrome is worse with C than with 5-FU. Activity and efficacy are similar. Capecitabine could be therefore considered standard of care in advanced CRC.
KeywordsAdvanced colorectal carcinomaCapecitabine5-FluourouracilSurvivalToxicity
Advanced non-resectable (metastatic) colorectal cancer (CRC) reported a dismal prognosis, and palliative chemotherapy is the only available oncologic treatment. 5-Fluorouracil (5-FU) remains an important agent in the treatment of advanced CRC. Mainly in the past, 5-FU was administered as a bolus injection either weekly or daily for 5 days, every 4–5 weeks. With these regimens, response rates (RRs) ranged approximately from 10 to 15%. The development of permanent venous access devices and portable infusion pumps has made the prolonged infusion of 5-FU possible on an outpatient basis. Capecitabine (C) is a fluoropyrimidine carbamate with antineoplastic activity. It is an orally administered systemic prodrug of 5′-deoxy-5-fluorouridine (5′-DFUR) converted to 5-fluorouracil, thus mimicking a continuos 5-FU infusion. Capecitabine is readily absorbed from the gastrointestinal tract. In the liver, a 60 kDa carboxyl esterase hydrolyzes much of the compound into 5′-deoxy-5-fluorocytidine (5′-DFCR). Subsequently, cytidine deaminase, an enzyme found in most tissues including tumors, converts 5′-DFCR into 5′-deoxy-5-fluorouridine (5′-DFUR). The enzyme thymidine phosphorylase (dThdPase) then hydrolyzes 5′-DFUR into the active drug 5-FU. Many tissues throughout the body express thymidine phosphorylase. Some human carcinomas express this enzyme in higher concentrations than the surrounding normal tissues. The rationale of the development of an oral fluoropyrimidine agent is to obtain an even higher concentration of 5-FU in neoplastic tissues. In a phase-III trial of previously untreated patients with metastatic colon cancer, C produced higher response rates than 5-FU and leucovorin, while the overall survival and time to disease progression were similar (non-inferior). As established in European trials, the recommended dose of C is 2,500 mg/m2 per day, administered twice a day for 14 days followed by a 1-week rest period. The side effects of C tend to be similar to those observed with prolonged infusion of 5-FU, with hand-foot syndrome being the most common. Although 5-FU remains the backbone of most regimens, the new agents irinotecan and oxaliplatin have become an important part of front-line treatment of this disease in the United States and abroad. The recent development of molecular targeted agents has provided additional improvements in both response and survival in patients with CRC. Standard first-line therapy is conventionally represented by fluoropyrimidine-based doublets or triplets, potentially combined with a biologic agent depending on the tumor characteristics, patients’ conditions and preferences, physician’s choice, and aim of treatment.
Colorectal 5-FU-based schedules naturally evolved into the substitution of 5-FU with C, thus avoiding the need for IV continuous infusion and the positioning of a central venous catheter. Oxaliplatin (XELOX) or irinotecan + C (XELIRI) has therefore become standard options in first or subsequent lines of advanced CRC patients. Up to now, there are several randomized studies confirming the non-inferiority of C-based doublets compared to conventional doublets, namely FOLFOX and FOLFIRI. Capecitabine and oxaliplatin may provide significant benefit. The results of completed clinical trials with C and oxaliplatin are similar to those obtained with FOLFOX. Also capecitabine and irinotecan may provide some benefit. However, this combination should be used with caution. Although some clinical trials have shown this combination to be tolerable and active, other trials with C and irinotecan have shown significant toxicity. The next step was the addition of a biologic agent (bevacizumab or cetuximab) to C-based doublets. Patients finally expressed a clear preference for oral drugs, which provided a highly positive influence on the acceptance of the treatment [30, 31].
The convenience of oral administration and the equivalent results in randomized trials with regard to standard continuous infusion 5-FU schemes prompted to the wide adoption of C as a standard option in patients with stage-IV CRC. In order to confirm the benefit of C-schedules (either as single agent, doublet, or combined with a targeted therapy), we have performed a pooled-analysis of all randomized, controlled trials (RCTs) comparing C with IV 5-FU either alone or combined with oxaliplatin or irinotecan plus or minus another biologic agent. Selected studies enrolled patients either in first- or subsequent-line settings. The aim was to confirm the efficacy of C-based treatment compared with conventional 5-FU formulation in terms of RR, progression-free and overall survival (OS).
A literature search was carried out to identify all relevant RCTs comparing C plus other agents (either chemotherapy or biologics) with 5-FU plus the same agents in the treatment of advanced CRC. Medline, EMBASE, and The Cochrane Library were used systematically in the search of all the articles published between January 2000 and October 2010 which included the following terms in their titles, abstracts, or keyword lists: colorectal cancer, C, metastatic, or advanced (limited to RCTs). Abstracts taken from various meetings of the American Society of Clinical Oncology (ASCO: http://www.asco.org) held between 2000 and 2010 were also systematically searched.
The reference lists of all the articles we found on this topic were examined manually. The quotations selected from this initial search were subsequently screened for eligibility by using the following criteria: (1) Patients with advanced (non-resectable, metastatic, or recurrent) CRC, (2) C plus chemotherapy ± a biologic agent versus 5-FU plus the same chemotherapy ± a biologic agent, (3) RCT, (4) report response rate and hazard ratios for progression-free survival (PFS) and OS. All abstracts, comments, reviews, and guidelines prior to the publication were excluded from the meta-analysis.
The following information were collected: author’s name or name of RCTs, publication year, number of patients, regimen, mean age, gender rate, Eastern Cooperative Oncology Group (ECOG) performance status or Karnofsky performance status, HR for OS, HR for PFS, and overall response rate (ORR).
A pooled-analysis of all RCTs comparing the efficacy of C plus chemotherapy (and potential targeted agents) and of 5-FU plus chemotherapy (and potential associated targeted agents) was performed. The endpoints used in this study were OS, defined as the time from random assignment to death from any cause censoring the patients who had not died at the date they were last known to be alive, PFS, defined as the time from random assignment to first documented progression or death, regardless of what has occurred before, ORR, defined as the sum of partial and complete response rates, according to the Response Evaluation Criteria in Solid Tumors. The overall HRs for OS and PFS were derived directly from the analysed article, while the ORR was calculated from published data (abstract or full paper). The method used is a pooled-analysis of published data only and not an individual patient data (IPD). With an adequate follow up and where events happen quickly and so publications report quite mature data, as in a population of metastatic CRC, a published data pooled-analysis for time-to-event outcome may obtain very similar results to an IPD meta-analysis. The main toxicities (as neutropoenia, febrile neutropoenia, nausea, vomiting, diarrhea, hand and foot syndrome, stomatitis, and cardiovascular ones) were also analyzed. A statistical test with a P-value < 0.05 was considered significant. Data analyses were performed on the intent-to-treat population, defined as all randomly assigned patients. RevMan 5.0.24 (Cochrane IMS) has been used for statistical analysis. For our analysis, we have used either a fixed-effect model (weighted with inverse variance) or a random-effect model. For each analysis (toxicity and efficacy), Cochran’s Q statistic and I2 statistics have been calculated first, in order to assess the heterogeneity among the proportions of the included trials. In case the P-value was found to be less than 0.1, the assumption of homogeneity was deemed invalid and the random-effect model was reported after exploring the causes of heterogeneity. Otherwise, the fixed-effect model was reported. A RR > 1 for response favored the control arms, while HRs < 1 for PFS and OS provided greater benefit with C-based chemotherapy versus no-C treatments.
The search strategy produced 19 studies (16 full texts and 3 abstracts) [1–18], all of which were RCTs. Only in 1 trial, a placebo arm versus an active arm (celecoxib) was used . Eight of them provided oxaliplatin schedules, six provided irinotecan schedules, and only two provided monotherapy arms (C vs. 5-FU). From these, two RCTs were subsequently published as a pooled-analysis and this final publication was obtained for the meta-analysis. Sixteen RCTs (14 full texts and 2 abstracts) were finally included in the pooled-analysis [1–6, 8–17]. One trial  was excluded because it reported the subgroup analysis (elderly data only) of another included trial. Two similar trials [6, 17] were also pooled for efficacy analysis 3 years later in a subsequent publication, and only the latest publication was included for PFS and OS . One trial  reported first the chemotherapy schedule comparison and then, in a separate publication, the results achieved after the addition of the monoclonal antibody bevacizumab to chemotherapy (this last analysis, that compared all chemotherapy schedules plus or minus bevacizumab, was not considered). One abstract was not considered in the meta-analysis either, due to the fact that the number of patients and the HRs for PFS and OS (for efficacy analysis) in each arm were not reported . All 16 trials were analyzed for overall response rate, 10 trials were analyzed for PFS, and 9 for the OS outcome.
characteristics of randomized trials analysed
N° pts/median age (y)
Control arm (n°)/exp. arm (n°)
Regimens control arm/exp. arm
ORR control arm/exp.arm (%)
PFS/TTP months control/exp. (HR)
OS months control/exp. (HR)
Median follow up (months)
FOLFOX-4 ± B/XELOX ± B
8.5 versus 8.0 (1.04)/6.3 vs 5.9 (1.08)
6.5 versus 6.6 (0.92)/4.7 versus 4.9 (0.92)
Diaz Rubio 
FUOX versus XELOX
6.9 versus 6.0 (1.15)/9.5 versus 8.9 (1.18)
144 (+57 B)/141 (+60 B)/145
FOLFIRI or mILF ± B/CapeIRI ± celecoxib/P
47.2% and 43.3%/38.6% 57.9%/53.3%a
NR (FOLFIRI versus CapIRI 1.36, mIFL versus CapIRI 1.05/NR 11.2 versus 8.3a/NR
99 (+141 B)/48 (+72 B)
FOLFOX-6 or bFOL/CapeOx ± B
41% and 20%/27% 52% and 39%/46%a
8.7 versus 6.9 versus 5.9 [9.9 vs 8.3 vs 10.3a](NR)/6.5 versus 4.9 versus 4.4 [5.8 vs 5.5 vs 5.5a](NR)
19.2 versus 17.9 versus 17.2 [26.1 vs 20.4 vs 24.6a]
19.9,15.1, 15 [17.9, 17.6, 18.5a]
3.1 versus 4.1 (0.90)/4.7 versus 4.3 (1.03)
FOLFIRI/CAPIRI ± celecoxib/P
7.8/6.9b (NR) 9.6/5.9c (NR)
19.9(0.8)/18.3(1)b 19.9 (0.31)/14.7 (1)c
NR/7.0 versus 9.0
8.0 versus 7.1 (1.17)/6.0 versus 5.1 (1.14)
4.8 versus 4.7 (0.97)/6.2 versus 5.6 (1.15)
Van Cutsem  apooled-analysis of ref 6,17
9.4 versus 8.1 (0.8)/4.7 versus 4.6 (0.997)
9.3 versus 8.8 (1)/6.8 versus 6.1 (1.32)
FOLFIRI-B versus CAPIRI-B
10.5 versus 10.1/NR
Van Cutsem 
4.7 versus 5.2
12.1 versus 13.2
Munoz Llarena 
Saltz versus XELIRI
The toxicities observed were those mainly related to 5-FU administration: diarrhea, nausea, vomiting, stomatitis, hand and foot syndrome, neutropoenia, febrile neutropoenia, and cardio toxicity.
Overall response rate
5-Fluorouracil is the cornerstone of treatment for early and advanced CRC. Continuous infusion represents the more active and less toxic form of administration of this cytotoxic drug; however, it involves the placement of a central venous catheter, thus increasing the risk of catheter-related thrombosis and infections. The development of an effective oral fluoropyrimidine represents a revolution in the treatment of cancer. Capecitabine, an oral pro-drug of 5-FU, currently represents a viable option as adjuvant treatment of CRC according to 2 randomized trials that confirmed the non-inferiority of C versus 5-FU bolus alone or with oxaliplatin [20, 21]. In advanced CRC, C, either alone or in combination with oxaliplatin, demonstrated non-inferiority compared to the standard formulation of 5-FU (either bolus 5-FU/folinic acid or FOLFOX4) in 3 landmark randomized trials [1, 13]. After 5-FU doublets have become standard first-line therapy for metastatic disease and early and advanced disease trials have been released, a plethora of randomized studies have been published, comparing C-based chemotherapy plus or minus biologic agents (namely bevacizumab or cetuximab) with conventional iv 5-FU doublets (FOLFOX or FOLFIRI). This pooled-analysis has pooled the results of all these randomized trials in order to confirm the efficacy of C-based regimens in terms of survival, response rate, and toxicity. All the trials were randomized and compared the same drugs (oxaliplatin or irinotecan plus or minus a biologic targeted therapy) plus either C or 5-FU, they were all first-line studies except for one, that enrolled patients in second-line setting. The results implemented another similar meta-analysis evaluating only C + oxaliplatin doublets published in the last months, whose articles have not been presented orally nor published in 2010. In particular, Ling’s meta-analysis reported similar outcome results but significant differences in toxicity for neutropoenia in favor of C . Our analysis, updated with more trials, confirms that diarrhea does not worsen under C regimens.
The toxicity results are impressive, as diarrhea is not significantly greater with C combinations, and have been confirmed even after the exclusion of irinotecan trials. Vomiting and nausea are the same in both the comparative arms. Stomatitis and neutropoenia are dramatically reduced with C-based chemotherapy (60% lower risk), which represents an excellent clinical result, considering that neutropoenia and mucositis are the most life-threatening complications of cytotoxic agents and also decrease the quality of life, since stomatitis hinders oral intake, favors infections and causes severe pain. In particular, severe stomatitis and neutropoenia are dramatically reduced by 69 and 74%. Febrile neutropoenia shows a trend toward reduced risk with C, but this is not significant. Finally, despite having been reported in 3 trials only, cardio toxicity is equal in both the comparison arms. Since hand-foot syndrome, a typical adverse event that affects palms and soles, develops at a high incidence (RR 3.45), the therapy needs to be properly managed in order to be successful. A recent Japanese study  shows that the incidence of grade-2/3 hand-foot syndrome was 21.1 and 75% in 39 and 8 patients, respectively, who were treated with supportive therapy (topical humectants and oral vitamin B6 preparation) from the start of drug administration and after several courses of drug administration or development of symptoms. This suggested the usefulness of early supportive therapy. A proper exchange of information between medical staff and patients may ensure the adequate management of adverse events and thus the achievement of better results from the therapy.
The primary aims for the acceptance of oral therapy are adherence and preference, other than toxicity. Discontinuation rates in C arms are low. The higher figures concern the first randomized C trial (Van Cutsem et al.), where the dose of C was 2500 mg/m2 per day (a poorly tolerated schedule in current clinical practice) and the rate of non-adherence was 20%, and the Comella et al. trial, where oxaliplatin and C administration was repeated every 2 weeks (instead of 3 weeks, as in classical Cassidy XELOX schedule). The administration of C is not underreported even in the elderly, where advanced age is not per se considered a major risk factor for inadequate compliance to therapy, at least according to a breast cancer adjuvant trial . Capecitabine is overall safe and convenient in advanced age and in the presence of hepatic and renal dysfunction, provided that the total dose is appropriately adjusted. In the last months, Jackson  published a subgroup analysis of elderly patients from Fuchs’ trial comparing FOLFIRI, modified IFL, and XELIRI. The only toxicities that significantly worsened in elderly population were asthenia and dehidratation, most likely due to diarrhea (55% in XELIRI vs. 11 and 23% in the other arms, although not significantly different). Drug holiday may help avoid toxicity exposure by stopping the administration of cytotoxic agents more frequently related to cumulative toxicity (e.g., oxaliplatin) and continuing the administration of the oral agent and the activity of a multidisciplinary caregiver team. A recent study  has demonstrated the safety and the prolonged efficacy of a maintenance C-therapy after an induction (4-cycles) XELOX regimen in non-progressing patients with advanced CRC. A significant reduction in the vast majority of all grades of adverse events occurred after switching from initial XELOX to maintenance C-chemotherapy, including grade 1,2, and 3 neuropathy, diarrhea, and lethargy. A median OS of more than 20 months is promising, and by limiting the number of oxaliplatin infusions, this approach minimizes the risk of unwanted cumulative neurotoxicity, is cheaper and more convenient for both patients and healthcare providers.
Provided that safety and toxicity are maintained, the goals of chemotherapy in advanced disease setting are activity and efficacy. Our analysis shows that RR, PFS, and OS are exactly the same in both C and 5-FU arms.
Standard first-line chemotherapy in suitable patients is a combination of fluoropyrimidine-based chemotherapy and either oxaliplatin or irinotecan plus or minus bevacizumab (or cetuximab, in case of KRAS wild-type CRC). This pooled-analysis confirms that C can safely replace 5-FU without compromising the efficacy of the treatment and that it should become the standard of care, also according to physicians’ preferences and patients’ attitudes and compliance . The addition of C to bevacizumab/oxaliplatin-containing schedules does not compromise the safety and the efficacy of the treatment, as the TREE2 trial demonstrates. Higher levels of gastrointestinal toxicity have been reported in 2 trials when bevacizumab was added to a combination of irinotecan plus C, although no differences were observed in terms of survival and RR. The addition of C to a cetuximab/oxaliplatin combination, however, seems to cause a high incidence of diarrhea, which is why the administration of C in the COIN trial has been significantly reduced . Thus, in case the implementation of C in schedules based on other drugs leads to overlapping toxicities, a dose adjustment of the oral drug will be required. Dosage reduction seems to be important also in elderly populations and in those with moderate renal impairment, especially in order to minimize the toxicity associated to C/irinotecan regimens. Although additional phase-3 studies are needed, C/irinotecan may be an effective regimen, allowing a greater number of treatment options for tumor control in patients with metastatic CRC.
The combination of C and bevacizumab has recently shown better results in PFS compared to C alone as first-line therapy in metastatic CRC, reaffirming the value of C when added to biologic agents other than cytotoxic ones .
Sequential phases of treatment represent an alternative strategy which consists in starting with the administration of a single agent and prescribing all the other drugs in sequence at progression of the disease. This strategy could be chosen for patients with asymptomatic, inoperable, and oligometastatic CRC that do not need an aggressive first-line therapy. A sequence of C → irinotecan → C/oxaliplatin or C/irinotecan → C/oxaliplatin has been tried and no differences have been reported in the overall median survival , except for a higher hand and foot incidence in the sequential arm, which may be related to a higher dose of C monotherapy.
Other oral 5-FU prodrugs have recently received some scientific attention. A randomized Japanese second-line trial comparing the FOLFIRI schedule with irinotecan/oral S1 (a combination of tegafur, 5-chloro-2,4-dihydroxypyridine, and potassium oxonate)  has demonstrated the non-inferiority (PFS) of the oral combination, and the worst toxicity reported was diarrhea in the irinotecan/S1 arm.
The major international guidelines (e.g., NCCN) now consider C ± oxaliplatin (±bevacizumab) as the first line of choice in advanced CRC patients and C + oxaliplatin as second-line treatment in irinotecan pretreated settings. Combinations of irinotecan and C (replacing FOLFIRI) need more phase-III trials in order to be regularly implemented as schedule of choice (alone or associated with targeted therapies). The first randomized trial comparing FOLFIRI + bevacizumab and XELIRI + bevacizumab enrolled 285 patients and was presented at the ASCO Meeting  in 2010. It is included in this meta-analysis for response and toxicity assessment. This trial showed no significant differences in terms of efficacy between XELIRI + bevacizumab and FOLFIRI + bevacizumab in patients with metastatic colorectal cancer treated in the first-line setting. However, the toxicity profile showed more neutropoenia and metabolic disorders in the FOLFIRI arm and more vomiting in the XELIRI arm. No differences were observed for diarrhea, maybe due to the reduction in the dose of irinotecan (240 mg/m2 vs. 250 mg/m2 as in other published trials).
The weakness of this pooled-analysis is that it derives outcome estimates (HRs) from published data only and not from IPD. Although the median follow up of all studies is appropriate for so type of population (18 months) and a meta-analysis of published data is a reasonable estimate of the HRs obtained from IPD, it cannot circumvent the other potential biases associated with this data (publication, patient exclusion, selective reporting, and follow-up bias). Despite these limitations, it can be affirmed undoubtedly, that C is as effective as and less toxic of iv 5-FU.
This pooled-analysis has demonstrated that the substitution of iv (bolus and continuous infusion) 5-FU with C does not compromise the efficacy of the treatment and significantly reduces the toxicity. The risk of stomatitis and neutropoenia, two dose-limiting toxicities of chemotherapy, is reduced by 60% with C in place of 5-FU. Despite the exclusion of 5-FU bolus (older) trials (a major cause of mucositis and leucopoenia), the result is a 50% lower risk of these toxicities. Diarrhea and cardiac toxicity are not increased, while vomiting is only slightly worse but easily controlled with standard antiemetic medication. Only hand and foot syndrome showed significant worsening, which can, however, be prevented and properly managed. Supportive measures for pain and discomfort reduction and for the prevention of secondary infections are also very important. Overall, C has shown to have the same efficacy as 5-FU, and therefore it is considered a valid substitute in standard chemotherapy schedules for advanced CRC (plus or minus a biologic agent except for anti-EGFR agents) and represents now the treatment of choice according to the major international guidelines. C has become the first choice of treatment also in other gastrointestinal neoplastic diseases such as gastric and pancreatic cancer. Toxicity and adherence to treatment need to be strictly monitored by treating only suitable and compliant patients and by reducing doses as appropriate (mainly in elderly patients). Trials on new orally available 5-FU prodrugs confirm that iv 5-FU may soon be forgotten, thus avoiding the need for central venous catheter.
In conclusion, C is the cornerstone of treatment for advanced (other than early) CRC because it is convenient, safe, and cost saving (toxicities, catheters, and outpatient clinic, for example) for both patients and the healthcare system.