Adjuvant chemoradiotherapy (CRT) improves relapse-free (RFS) and overall survival (OS) in patients with resected gastric cancer. However, difficulties in standardizing an optimal surgical approach and a perceived higher toxicity compared with the perioperative approach have limited its widespread application in Europe. The aim of our study was to assess toxicity and long-term outcomes of adjuvant CRT at our institution.
A retrospective review (September 2001-January 2012) was completed of patients with resected gastric cancer who received adjuvant CRT (Macdonald regimen). Adverse events and completion rates, RFS and OS were estimated. Univariate and multivariate analyses of prognostic factors for OS were performed.
Eighty-seven patients were included. Most had diffuse (52%) and locally advanced tumors (stage III-IV; 66.7%). D2 lymphadenectomy was performed in 80.5%. The most frequent grade 3-4 toxicities were gastrointestinal (28%) and stomatitis (20%), with 78.2% completing treatment. With a median follow-up of 115 months, 58.5% had relapsed, most of them distantly. Median RFS and OS were 9 and 24 months, respectively. Univariate analysis showed that performance status, stage and lymph node burden were significant factors for OS. In the multivariate study, only stage and lymph node burden remained as independent OS predictors.
Our implementation of the Macdonald regimen achieved worse outcomes than those reported in the INT-0116 trial. The rate of distant relapse remains unacceptably high. Higher rate of positive lymph nodes and of diffuse tumors could explain some differences. The use of perioperative chemotherapy, especially in patients with a poorer prognosis, might improve these results.
Tumori 2015; 101(5): 517 - 523
Article Type: ORIGINAL RESEARCH ARTICLE
AuthorsGema Bruixola, Ángel Segura, Robert Díaz-Beveridge, Javier Caballero, Mohamed Hassan Bennis, Laura Palomar, Alejandra Giménez, Fernando Mingol, Carmen García-Mora, Jorge Aparicio
- • Accepted on 09/04/2015
- • Available online on 28/05/2015
- • Published in print on 09/09/2015
This article is available as full text PDF.
Gastric adenocarcinoma is the second most common cause of cancer-related death worldwide (1). A radical surgery with adequate margins is the preferred treatment for those patients with localized disease (2); however, the optimal type of lymph node resection is still unclear, and practice varies between Western and Asian patients (3-4-5-6). Unfortunately, despite surgical improvements, the outcome in patients who undergo a curative gastrectomy for locally advanced tumors is poor, with a 5-year survival rate of less than 35% (7). In those patients with unresectable or metastatic disease, long-term survival remains anecdotal.
The potential benefit of adjuvant treatment alongside surgery to improve these dismal results has been explored in several trials, with initial negative results (8-9-10-11). However, in the last 15 years, 2 different adjuvant approaches have achieved a positive impact on survival in this setting. The SWOG Intergroup Study 0116 (INT-0115), a large randomized phase III trial (12) reported a significant overall survival (OS) benefit with the use of postoperative adjuvant chemoradiotherapy (CRT; 5-FU and leucovorin and radiotherapy) compared with surgery alone. The MAGIC trial (13), which evaluated perioperative adjuvant chemotherapy without radiotherapy with the ECF regimen (epirubicin + cisplatin + 5-fluorouracil), also showed a benefit both in terms of OS and recurrence-free survival (RFS), compared with a surgery-only approach. No trial has compared head-to-head both adjuvant approaches, and practice varies widely, with the Macdonald CRT regimen (12) favored in the United States, while the perioperative strategy is used extensively in Europe (13).
In the present study, we aimed to evaluate the toxicity, completion rate, patterns of recurrence and survival of the Macdonald CRT regimen, which was implemented at our institution in daily clinical practice in 2001.
Materials and methods
We performed a retrospective review of all consecutive patients with pathologically confirmed gastroesophageal junction or gastric adenocarcinoma who underwent surgery with curative intent and were treated with adjuvant CRT at our institution. The study was approved by the local Ethics Committee of Clinical Research and was conducted according to the Declaration of Helsinki for studies in humans.
All patients treated with adjuvant CRT between September 2001 and January 2012 were identified using our prospectively driven clinical databases. Patients eligible for adjuvant CRT had an Eastern Cooperative Oncology Group (ECOG) performance status (PS) ≤2, T3-4 and/or N+ gastric cancer and adequate hepatic, renal and bone marrow function. No preoperative chemotherapy was permitted. Patients with metastatic disease were excluded from the analysis.
Clinical data collected included demographics, histological subtype, primary surgery outcome and type of lymph node dissection, timing of radiotherapy, chemotherapy dose adjustments, acute toxicity, complication and completion rates, reasons for discontinuation, time and site of first relapse, and survival. The tumor was staged according to the 6th edition of the American Joint Committee on Cancer (AJCC) staging manual (14).
The Macdonald regimen consisted of 1 cycle of (5-FU; 425 mg/m2 per day) and leucovorin (20 mg/m2 per day) daily for 5 days, followed 1 month later by 45 Gy (1.8 Gy/day) radiotherapy (RT) given with 5-FU and leucovorin (400 mg/m2 and 20 mg/m2, respectively). On days 1-4 and on days 23-25. RT was administered according to the INT-0116 protocol. Two further 5-day cycles of chemotherapy (5-FU 425 mg/m2 per day and leucovorin 20 mg/m2 per day) were given at monthly intervals beginning 1 month after completion of RT.
Toxicity was graded using the NCI Common Terminology Criteria for Adverse Events (CTCAE) version 4.3.
RFS was measured from the date of primary radical surgery to the date of first recurrence of disease, while OS was measured from date of first surgery to the date of death or date of last contact.
Sites of first failure (i.e., locoregional or distant) were also collected. Locoregional recurrence was defined as any recurrence in the tumor bed, anastomotic site, gastric remnant, duodenal stump or regional nodes within the irradiated volume. Distant metastases were defined as any recurrence outside of the irradiated field, including metastases to the liver, lower para-aortic lymph nodes, extra-abdominal sites and peritoneal seeding.
The Kaplan-Meier method was used to compute RFS and OS. Summary statistics were calculated for continuous and categorical variables. Differences were compared using Cox regression models. Data were analyzed by means of the SPSS 21 package (IBM, Armonk, NY, USA). A p value of less than 0.05 was considered statistically significant.
Eighty-seven patients were identified for inclusion in the study. Their baseline characteristics and surgical procedures performed are shown in
Clinical and surgery baseline data (N = 87)
|Characteristics||Patients, no. (%)|
|ECOG PS = Eastern Cooperative Oncology Group performance status.|
|Median age, years (range)||59 (35-79)|
|6th Edition TNM staging|
|IVA (nonmetastatic)||21 (24.1%)|
|Linitis plastic||25 (28.4%)|
|Adjuvant treatment indication|
Sixty-eight patients (78.2%) completed treatment as planned. The full course of radiation therapy (4,500 cGy) was given in 85 patients (97.7%); only 2 patients stopped radiotherapy due to stomatitis and hematological toxicity, receiving a total dose of 4,100 cGy and 1,600 cGy, respectively; both discontinued chemotherapy for the same reason. Regarding chemotherapy, 19 patients (22%) did not receive all of the programmed cycles. In 15% of patients, a dose reduction of 25% was needed. One-week delays were necessary in 30% of patients. The most common cause of chemotherapy discontinuation was unacceptable toxicity in 15% of patients.
Treatment-related grade 3-4 adverse events (AEs) were experienced in 40% of patients. The most common was gastrointestinal toxicity in 24 patients (28%), with grade 3-4 diarrhea reported in 17 patients (19.5%) and emesis seen in 7 patients (8%). Twenty-two patients (25.3%) lost 10% or more of their body weight, and grade 3 stomatitis was observed in 17 patients (19.5%). Total parenteral nutrition was needed in 4 patients (4.6%). Grade 3 fatigue was reported in 6 patients (6.9%). Hospital admission was needed in 20 patients (23%). Hematological toxicities were rare, with only 5 patients with febrile neutropenia (5.6%) and 2 patients with grade 3 thrombocytopenia (2.3%). However, there were 2 treatment-related deaths (2.3%) secondary to sepsis in the context of a febrile neutropenia (both toxic deaths occurred during the first period of treatment, between 2001 and 2005).
With a median follow-up of 115 months (range 37-161), median time of RFS was 9.2 months (95% confidence interval [95% CI], 7.7-10.9) (
Recurrence-free survival (RFS).
Overall survival (OS).
The univariate analysis of prognostic factors showed that PS = 0, stage of disease <III and positive lymph node <10 were statistically significant factors for an improved OS. However sex, histology, weight loss, albumin and basal serum tumor marker levels were not significant in this analysis.
The number of positive lymph nodes had a statistically significant impact in OS; median OS times in patients with metastatic lymph nodes <5, 5-10, >10-15, >15-20 and >20 were 74.2, 33.35, 11.3, 11.12 and 9.3 months, respectively (p<0.0001). While not statistically significant, patients who underwent a D2 lymphadenectomy had a better median survival (31.7 months; 95% CI, 17.2-46.1) than those who underwent a D1 or D0 lymphadenectomy (22.5 and 7.8 months, respectively) (p = 0.051).
Asymptomatic patients with PS = 0 had a better median OS rate (33.7 months; 95% CI, 0-72.8), than those with PS = 1 (24.2 months; 95% CI, 16.3-32.3); this was statistically significant (p<0.0001).
There was a trend for patients with diffuse tumors to have worse median OS (20.5 months; 95% CI, 12.9-28.2) compared with those patients with intestinal tumors (median OS: 51.9 months; 95% CI, 20.9-82) (p = 0.057).
After stratifying according to the treatment period (2001-2005 vs. 2006-2012), an improvement in survival in patients treated in the second period (median OS: 33 months; 95% CI, 4-63) versus the first period (median OS: 20.9 months; 95% CI, 17.2-24.6) could be seen. However this was not statistically significant.
For the multivariate analysis, we used the 3 statistically significant variables. Stage of disease III or higher after surgery (p = 0.001; hazard ratio [HR] = 2.017; 95% CI, 1.57-2.58) and a positive lymph node count >10 p = 0.001; HR = 19.761; 95% CI, 5.22-74.64) remained independent factors associated with a poor baseline prognosis, whereas PS did not (
Multivariate analyses of OS according to clinicopathological variables
|Variables||HR||95% CI||p Value|
|CI = confidence interval; ECOG PS = Eastern Cooperative Oncology Group performance status; HR = hazard ratio; OS = overall survival.|
|ECOG PS 0 vs. 1||1.73||0.96-3.09||0.065|
|Clinical tumor stage <III vs. >=III||2.017||1.57-2.58||0.001|
|Lymph node burden <10 vs. >10||19||5.232-74.64||0.001|
Patterns of recurrence
Fifty-one patients (48.5%) relapsed during the follow-up period.
Patterns of recurrence (N = 87)
|Patterns of recurrence||No. (%)|
|*A patient could have more than 1 metastatic site.|
|Locoregional recurrence||1 (1.1%)|
|Distant recurrence||45 (51.7%)|
|Locoregional and distant*||5 (5.7%)|
|Sites of distant recurrence||50 (57.4%)|
|Lymph nodes||4 (4.5%)|
After relapse, 31 patients (35.2%) received first-line palliative chemotherapy, 6 (6.9%) received 2 lines of chemotherapy, 1 (1.1%) was treated with palliative radiotherapy and 19 patients (21.5%) were only candidates for best supportive care.
After the publication of the INT-0116 trial that showed a clear survival benefit with the use of adjuvant CRT in R0-resected gastric cancer patients, the Macdonald regimen became a new standard of treatment (12). The recent analysis by Smalley et al (15) with a longer follow-up, has confirmed that the benefit in OS and RFS remains highly significant in the long-term. However, concerns about potential serious toxicity with the regimen and the suboptimal surgical management in the study (with only 10% of patients undergoing a D2 lymphadenectomy) have made many physicians reluctant to use this treatment approach.
On the basis of these published results and the level 1 recommendation, we decided to implement the Macdonald regimen in our routine daily practice from September 2001. However, in a real-life setting, the selection of patient candidates for this treatment as well as the management of toxicity may differ from that seen in clinical trials.
Comparison between the Intergroup Study 0116 (INT-0116) and recent studies with the Macdonald regimen
|INT-0116 (2001) (12)||Current study||Aftimos et al (2010) (16)||Kundel et al (2011) (17)||Kim et al (2011) (18)||Misirlioglu et al(2014) (19)|
|CRT = chemoradiotherapy; NR = not reported; OS = overall survival; RFS = relapse-free survival; RT = radiotherapy.|
|Median age, years (range)||60 (25-87)||59 (35-79)||58.3 (35-80)||63 (23-86)||56 (23-71)||52 (20-78)|
|Patients CRT (Macdonald)||281||87||14||166||80||185|
|T3-T4||68%||21%||75% (only T3)||77%||76%||67%|
|1-3||42%||25%||46%||48% N1||24% N1||NR|
|≥4||43%||53%||25%||22% N2 15% N3||55% N2 14% N3||NR|
|Median follow-up, months (range)||10.3 years||115 (37-161)||11.5 (range NR)||51 (2-112)||48 ( 3-83)||99 (79-132)|
|CRT completed (%)||64%||78.2%||36.8%||54%||97%||NR|
|RT completed (%)||NR||97.7%||91.7%||87%||97%||88%|
|Toxic deaths, no. (%)||3 (1%)||2 (2%)||4 (16.6%)||3 (1.88%)||0||0|
|Gastrointestinal toxicity (G3-4)||33%||28%||36%||29%||11%||26%|
|Hematological toxicity (G3-4)||54%||8%||40%||35%||5%||8%|
|RFS, months||3||9.2||10.5||R0 not reached||Not reached||NR|
|OS, months||36||24.3||75||R0 not reached||Not reached||NR|
|Distant relapses, no. (%)||40 (33%)||50 (57%)||2 (11%)||60%||38 (36%)||NR|
Our CRT completion rate of 78.2% was encouraging, since it was higher than the ones reported both in the original INT0016 trial (64%) (12) and in most series, where they have ranged from 36% to 54% (15, 16, 18). As could be expected, this was associated with our better toxicity profile, which also implied fewer discontinuations of treatment. Major toxicities observed were hematological and gastrointestinal. Remarkably, our rate of hematological toxicity was only 7.9%, lower than those reported in older series, which ranged from 35% to 54% (12, 15, 16), and which follows the trend of more recent series with reported rates ranging from 5% to 8% (17, 18). However, there were 2 toxic deaths due to sepsis during febrile neutropenia. In contrast, gastrointestinal toxicity was observed in 28% of our patients, well within the normal range of most studies (26%-36%) (12, 15, 16, 18), with the exception of the 11% rate described by Kim et al (18). In the present study, the rate of hospital admissions was 23%, less than in other published data (16). The goal of several of the admissions in our series was to provide intensive supportive treatment, to avoid treatment discontinuations; this might also explain the differences observed in hematological toxicities.
Despite the higher completion rate and the acceptable toxicity profile, however, our RFS and OS results were inferior to those presented in the INT-0116 trial (12-13-14) studies. Differences in baseline prognostic factors may account for some of these worse results, as our population included a significant percentage of patients with a poorer prognosis. Most patients were symptomatic (PS = 1 in 57.4% of patients), had diffuse tumors (52% of patients) with a high lymph node burden (78.2% with N+, but 52.9% with more than 4 metastatic lymph nodes). Noteworthy is the fact that almost a third of patients had linitis plastica, a well-known poor prognostic factor. Our data are consistent with those of a single-institution experience published recently by Misirlioglu et al (19), which has reported similar 1-, 3- and 5-year OS rates; it supports the suggestion that outside the framework of clinical trials, it could be difficult to reach these results.
Although it did not reach statistical significance, the high proportion of patients with diffuse histology could have played a key role in our poor survival outcomes; it is known from the Macdonald study that this subgroup of patients benefit less from adjuvant therapy, for reasons that remain unclear. Nevertheless, diffuse histology is underrepresented in the study of Misirlioglu et al (19), and yet despite that, their published survival data are comparable to ours. In this sense, the impact of histology on survival is still unclear, but we think it warrants further studies.
The role of RT in patients with extended radical surgery is controversial, as the risk of local relapse is smaller. In our series, although patients with D2 lymphadenectomies had improved survival rates, this was not statistically significant, probably due to the small sample size. A recent meta-analysis (20) of 6 studies with 2,135 patients, concluded counterintuitively that postoperative adjuvant CRT may be associated with a longer 5-year OS and 5-year RFS in patients with a D2 lymphadenectomy. It is also worth noting that 2 trials have shown a clear OS benefit with the use of adjuvant chemotherapy alone (i.e., S-1 (21) or capecitabine-oxaliplatin (22)) in Asian patients with radically resected gastric cancers with extended lymphadenectomies. In this sense, future prospective studies are needed to determine the role of CRT in patients with D2 dissections (23, 24) and the role of adjuvant chemotherapy in Western patients.
Our study also showed a trend for a higher survival rate in patients that had been treated in the second period of time (2006-2012); the learning curve of the surgical teams and technological improvements in RT techniques can explain these differences, which have been also described in other series (24, 25).
Finally, as would be expected, a higher stage at diagnosis and a higher lymph node burden were significantly independent poor-prognosis factors. Of note, the local relapse rate was low in our series (6.8% of patients), and most patients died of metastatic disease. These patients might benefit preferentially from the use of perioperative or adjuvant higher-intensity chemotherapy, as in the previously mentioned Asian trials (24, 25), to decrease the risk of systemic recurrence, especially in the setting of a radical surgical procedure, where the risk of local relapse would presumably be low.
Our data support the use of the Macdonald regimen in a real-life setting as feasible, with a manageable safety profile with intensive supportive care that allows reasonable treatment completion rates. However, our long-term survival results were poorer than those of the INT-0116 trial (12, 14) and other series (15-16-17). A high local lymph node burden and a predominance of diffuse tumors may account for these differences. Improved surgical expertise and technical improvements in radiotherapy have decreased the risk of local relapse, but the rates of distant relapse remain unacceptably high. The use of higher-intensity perioperative chemotherapy, especially in patients with a poorer prognosis, might improve these results. The role of RT in those patients treated with intensive chemotherapy, especially after radical surgical resections, and how to combine RT with other, more complex chemotherapy regimens, remain unclear.
- Bruixola, Gema [PubMed] [Google Scholar] 1, * Corresponding Author (email@example.com)
- Segura, Ángel [PubMed] [Google Scholar] 1
- Díaz-Beveridge, Robert [PubMed] [Google Scholar] 1
- Caballero, Javier [PubMed] [Google Scholar] 1
- Bennis, Mohamed Hassan [PubMed] [Google Scholar] 1
- Palomar, Laura [PubMed] [Google Scholar] 1
- Giménez, Alejandra [PubMed] [Google Scholar] 1
- Mingol, Fernando [PubMed] [Google Scholar] 2
- García-Mora, Carmen [PubMed] [Google Scholar] 3
- Aparicio, Jorge [PubMed] [Google Scholar] 1
Department of Medical Oncology, La Fe University and Polytechnic Hospital, Valencia - Spain
Department of General and Digestive Surgery, La Fe University and Polytechnic Hospital, Valencia - Spain
Department of Radiation Oncology, La Fe University and Polytechnic Hospital, Valencia - Spain