Perioperative chemotherapy for gastroesophageal cancer in British Columbia : a multicentre experience

The BC Cancer Agency (BCCA) is a multicentre institution that treats most oncology patients for the province. Characteristics of the 83 BCCA patients with localized gastric, gastroesophageal junction, or lower esophageal cancer who initiated perioperative chemotherapy either ECF or epirubicin, cisplatin, and capecitabine (ECX) from 2008 to 2011 were abstracted to an anonymous database and analyzed.


INTRODUCTION
In Western populations, cancers originating in the esophagus, gastroesophageal junction (gej), and stomach represent a major health problem and are considered highly lethal diseases, with an overall 5-year mortality rate that ranges from 17% to 27% 1 .Unfortunately, most of these patients present with latestage disease, when curative therapy is not possible 1 .
Although surgical resection remains the only potentially curative treatment for nonmetastatic gastroesophageal cancer, surgery alone is associated with only a modest 5-year overall survival (os) rate of about 25%-35% [2][3][4][5] .Several studies have evaluated the roles of adjuvant chemotherapy or radiotherapy (or both) after curative-intent surgery for gastroesophageal tumours in the hope that better outcomes could be achieved.The most well-known study in the adjuvant setting is the INT0116 trial, in which chemoradiotherapy after complete resection (compared with surgery alone) demonstrated a significant survival benefit (36 months vs. 27 months, p = 0.005), leading to adoption of that particular regimen in the United States 6 .The positive results of that study have nevertheless been criticized by some because of the poor extent of lymphadenectomy, which might have

Background
In 2006, perioperative epirubicin, cisplatin, and 5-fluorouracil (ecf), compared with surgery alone, demonstrated a significant survival benefit in resectable gastroesophageal cancers.We report the results of our experience with that protocol.

Methods
The BC Cancer Agency (bcca) is a multicentre institution that treats most oncology patients for the province.Characteristics of the 83 bcca patients with localized gastric, gastroesophageal junction, or lower esophageal cancer who initiated perioperative chemotherapy either ecf or epirubicin, cisplatin, and capecitabine (ecx) from 2008 to 2011 were abstracted to an anonymous database and analyzed.

Results
Of the 83 patients in the cohort [66 men; median age: 62 years (range: 37-79 years)], 87.9% completed 3 cycles of perioperative chemotherapy, and 93.9% (n = 78) under went an attempt at surgery (2 patients died of chemotherapy toxicities, 1 refused surgery, and 2 developed disease progression before surgery).In 11 of the surgeries (14.1%), tumours could not be resected because of unresectability (n = 1), liver metastasis (n = 1), and peritoneal carcinomatosis (n = 9).One patient died of surgical complications.The 6 patients (7.2%) who achieved a pathologic complete response are all alive and recurrence-free.Of 46 patients (55.4%) who subsequently began postoperative chemotherapy, 44.5% completed 3 cycles.Estimated median survival was 40.3 months.Weight loss was the only significant prognostic factor for worse overall survival.

Current OnCOlOgy-VOlume 21, number 2, April 2014
Copyright © 2014 Multimed Inc.Following publication in Current Oncology, the full text of each article is available immediately and archived in PubMed Central (PMC).
led to an overestimation of the adjuvant chemoradiation benefit observed.Several other studies have also investigated the role of adjuvant chemotherapy after curative resection, but most have failed to demonstrate any improvement in os or recurrence-free survival (rfs) in Western populations [7][8][9][10][11] , prompting the evaluation of neoadjuvant approaches for locally advanced gastric cancer.Moreover, according to the literature, complete resection (R0) is achieved only in approximately 70% of patients who undergo gastroesophageal cancer surgery 2,3,12,13 .Neoadjuvant chemotherapy with or without radiation therapy has therefore recently been added to the surgical protocol, with the aim of downstaging tumours and improving the rates of R0 resection and survival [14][15][16] .
The landmark study comparing perioperative epirubicin, cisplatin, and 5-fluorouracil (ecf) with surgery alone in patients with resectable gastroesophageal cancer (the magic trial) was published in 2006.It demonstrated a significant os benefit in favour of the combination arm 14 .Since then, perioperative chemotherapy with either ecf or epirubicin, cisplatin, and capecitabine (ecx) has become the standard of care for resectable gastroesophageal cancer in British Columbia.More recently, the cross trial demonstrated that, compared with surgery alone, neoadjuvant chemoradiotherapy with weekly carboplatin and paclitaxel followed by surgery also increases os for patients with esophageal or gej tumours 16 .The option of neoadjuvant chemoradiation with weekly carboplatin and paclitaxel has therefore also been available for locally advanced esophageal or gej tumours at our centre since 2012.After incorporation of perioperative ecx or ecf for gastroesophageal tumours into our clinical practice, a determination of whether our results are comparable to those obtained in a strictly controlled clinical trial was extremely relevant.Our retrospective study was conducted under that premise.
The aims of the study were to investigate whether the results of the magic trial could be replicated in our non-experimental setting and to explore prognostic variables associated with better os in our patient population.

METHODS
The BC Cancer Agency is a multicentre institution that treats most oncology patients for the province of British Columbia.All patients with localized gastric, gej, or lower esophageal cancer who initiated ecx or ecf perioperative chemotherapy from March 2008 to June 2011 at our institution were identified using the pharmacy database.Patients with metastatic disease that was identified before chemotherapy commenced or who received radiation therapy as part of perioperative treatment were excluded.Baseline demographics, tumour characteristics, and treatment details were abstracted to an anonymous database and analyzed.This study was approved by the local Institutional Review Board.
Statistical analysis was performed using SPSS for Windows (version 14.0: SPSS, Chicago, IL, U.S.A.).Overall survival was calculated in months from the time of primary diagnosis to the date of death or last follow-up, and rfs was calculated from the time of primary diagnosis to the date of disease recurrence, death, or last follow-up.Kaplan-Meier curves for rfs and os were generated.The log-rank test was used to assess statistical differences between variables, with a p value less than 0.05 being considered statistically significant.Multivariable survival analyses using Cox proportional hazards models explored the effect of variables on os.Hazard ratios (hrs) and 95% confidence intervals were calculated to estimate risk of death.

Treatment
Of the 83 patients, 73 (87.9%) completed 3 preoperative cycles of either ecx or ecf.The response rate among patients who underwent imaging before surgery was 49.3% (response defined by the radiologist).Chemotherapy toxicities caused the death of 2 patients from febrile neutropenia, 1 patient refused surgery, and 2 patients developed disease progression before surgery.Surgery was attempted in the remaining 78 patients (93.9%), but in 11 of the surgeries (14.1%), the tumour could not be resected because of unresectability (n = 1), liver metastasis (n = 1), or peritoneal carcinomatosis (n = 9).Only 1 patient died of surgical complications.In 59 patients (71%),

Current OnCOlOgy-VOlume 21, number 2, April 2014
Copyright © 2014 Multimed Inc.Following publication in Current Oncology, the full text of each article is available immediately and archived in PubMed Central (PMC).complete resection (R0) was achieved.In those 59 patients, the median number of lymph nodes examined was 10 (range: 2-41), and the median number of lymph nodes with metastatic involvement was 1 (range: 0-21).The 6 patients who achieved a pathologic complete response were all alive and recurrence-free at the time of writing.Interestingly, all 6 patients with a pathologic complete response received 3 cycles of preoperative chemotherapy, and in imaging before surgery, 5 showed a marked response to that chemotherapy.(In 1 patient, no imaging was performed to evaluate response.) Among patients who achieved an R0 resection, a trend toward worse os was evident in the group having 3 or more lymph nodes with metastatic involvement than in the group having 2 or fewer lymph nodes with metastatic involvement (p = 0.23).Compared with patients in whom 10 or more lymph nodes were examined, patients having fewer than 10 lymph nodes resected also showed a trend toward worse os (p = 0.24).Overall survival tended to be better in patients who received 3 preoperative cycles of chemotherapy than in patients who could not complete the 3 planned cycles (p = 0.24).
On multivariate analysis, no statistical significance was observed for number of lymph nodes involved (hr: 1.13; p = 0.058), number of lymph nodes examined (hr: 0.95; p = 0.164), number of preoperative chemotherapy cycles (hr: 0.80; p = 0.62), or number of postoperative cycles (hr: 0.80; p = 0.29) in the cohort of patients who received a curative resection.However, a trend for worse os was again observed depending on the number of involved lymph nodes.The median os time for patients who achieved an R0 resection has not yet been reached (Figure 1).
At December 2012, 39 patients (47%) had died.The estimated median os for the entire cohort was 40.3 months (Figure 2).On univariate analyses, initial presentation with weight loss was associated with worse os (p < 0.001), the median os being 17.6 months for those who presented with a weight loss of at least 5 kg and not yet reached for the patients with no history of weight loss (Figure 3).Age, sex, prior history of smoking, comorbidities, number of preoperative chemotherapy cycles, and number of postoperative chemotherapy cycles were not independent prognostic factors for os.On multivariate analyses, only presentation with weight loss was significantly associated with worse os (hr: 0.196; p < 0.001).

DISCUSSION
Comparing the results of our retrospective study with those of the magic trial (Table ii), we observed similarities in terms both of median age at diagnosis and of male predominance.However, we also observed an important discrepancy with respect to tumour location: esophageal and gej tumours constituted a much higher proportion of the disease in our patients.That finding is not surprising because the magic trial initially included stomach cancers only; in 1999, the protocol was modified to include adenocarcinomas of the esophagus.Another important consideration is that all perioperative chemotherapy administered in the magic trial consisted of ecf.Given the noninferiority with ecx in the metastatic setting and the easier administration in that regimen, we incorporated ecx into the perioperative scenario as an alternative 17 .Therefore, compared with 100% of patients in the magic trial, only 27.7% of our patients received perioperative ecf.Despite the difference in the chemotherapy combinations used, the proportion of patients who completed 3 preoperative and 3 postoperative cycles were similar in both studies, as were the proportions of patients who underwent surgery and who achieved an R0 resection.
It is noteworthy that our estimated median os is 40.3 months, when in the magic trial, it was only about 26 months (estimated from the Kaplan-Meier curve because the relevant data were not provided).One potential reason for the observed difference might be the use of pet-ct imaging for staging in 67.5% of our cases; pet-ct was not performed in the magic trial.Accuracy in preoperative staging is known to be higher with pet-ct than with ct imaging (68% vs. 53%); pet-ct also identifies more clinically occult metastatic disease 18 .The selection bias implied by preventing more patients with incurable disease from receiving perioperative treatment could explain the improved os seen in our cohort.Perhaps more importantly, a high proportion of our patients received ecx (72.3%) instead of ecf.At least in the metastatic setting, the real-2 trial showed noninferiority for ecx compared with ecf 17 , and a recent meta-analysis demonstrated superior os for patients treated with capecitabine combinations compared with patients  receiving 5-fluorouracil combinations 19 .Whether capecitabine use also translates into better os for nonmetastatic disease remains unknown.The inclusion of a higher proportion of patients with esophageal and gej tumours might be another potential explanation for the better os in our study; however, univariate and multivariate analyses showed no difference in os according to tumour location.Lastly, the time period of the studies might also represent an advantage for our cohort, because staging and surgical procedures tend to improve over time.
Although lymph node dissection remains controversial, a recent study in 1377 patients showed that lymphadenectomy leads to improved outcomes 20 .The magic trial reported a 42.5% rate of D2 lymphadenectomy without mentioning the number of lymph nodes examined.By contrast, we could not determine the type of lymphadenectomy performed, but our median number of lymph nodes examined was 10.
In our study, the only statistically significant prognostic factor for worse os was weight loss, which was also previously reported by other authors [21][22][23][24][25] .For the subset of patients who achieved R0 resection, a trend toward worse os was observed for the group with 3 or more lymph nodes having metastatic involvement.Since 1999, response to neoadjuvant chemotherapy has been well established to be predictive of survival in patients with resectable gastric cancer 26,27 .Another more recent study conducted at Memorial Sloan-Kettering Cancer Center showed that 3-year disease-specific survival was significantly higher for patients achieving a better than 50% pathologic response to preoperative chemotherapy than for those achieving a lesser histologic response (69% vs. 44%) 28 .Currently, identifying the patients who would best respond to neoadjuvant therapy remains a challenge; no demographic variable or tumour characteristic can, as yet, identify responders a priori.
In the best-case scenario, preoperative treatment can induce a complete pathologic response, which is well known to be associated with improved outcomes in numerous malignancies, such as those of breast 29-32 , lung 33 , rectum 34 , and esophagus [35][36][37][38] .However, the prognostic value of a pathologic complete response after neoadjuvant treatment for gastric cancer is still a matter of debate 39 and deserves further analyses.Unfortunately, the magic trial has not reported a rate of complete pathologic response.In our cohort of patients, all who achieved a complete histologic response (7.2%) remained alive and had not experienced disease recurrence after a median follow-up of 40.3 months.That finding emphasizes the importance of complete pathologic response as a prognostic factor after administration of neoadjuvant chemotherapy and raises the question of whether intensifying preoperative chemotherapy could translate into higher rates of cure.
Our study has some limitations inherent to all retrospective analyses, especially the potential for selection bias associated with information obtained from chart reviews.Given its retrospective nature, our study could not determine whether the surgery that was performed included a D1 or D2 lymphadenectomy, because that description was not always available in the final pathology report.Moreover, the adverse events related to perioperative chemotherapy, which would otherwise increase the strength of the present study, could not be reviewed.

CONCLUSIONS
In summary, our multicentre experience confirms the feasibility of the magic protocol in the real-world context and shows that ecx is also an adequate regimen in the perioperative setting.Initial presentation with weight loss was the only significant prognostic factor for worse os.All patients who achieved a pathologic complete response had received 3 cycles of preoperative chemotherapy and were recurrence-free at the time of writing.Given the prognostic value of a complete response after neoadjuvant therapy, future studies should explore the utility of intensifying preoperative chemotherapy to determine whether intensification might contribute to even better outcomes.

figure 3
figure 3 Kaplan-Meir curve for overall survival in patients with weight loss at presentation.

figure 2
figure 2 Kaplan-Meir curve for overall survival in the entire study cohort.

table i
Demographic, clinical, tumour, and treatment characteristics Current OnCOlOgy-VOlume 21, number 2, April 2014 Copyright © 2014 Multimed Inc.Following publication in Current Oncology, the full text of each article is available immediately and archived in PubMed Central (PMC).

table ii
Comparison with the perioperative group from the magic trial .Wolmark N, Wang J, Mamounas E, Bryant J, Fisher B. Preoperative chemotherapy in patients with operable breast cancer: nine-year results from National Surgical Adjuvant Breast and Bowel Project B-18.J Natl Cancer Inst Monogr 2001;:96-102.30.Symmans WF, Peintinger F, Hatzis C, et al.Measurement of residual breast cancer burden to predict survival after neoadjuvant chemotherapy.J Clin Oncol 2007;25:4414-22.31.Adams S, Chakravarthy AB, Donach M, et al.Preoperative concurrent paclitaxel-radiation in locally advanced breast cancer: pathologic response correlates with five-year overall survival.Breast Cancer Res Treat 2010;124:723-32.32.Chavez-Macgregor M, Litton J, Chen H, et al.Pathologic complete response in breast cancer patients receiving anthracycline-and taxane-based neoadjuvant chemotherapy: evaluating the effect of race/ethnicity.Cancer 2010;116:4168-77.33.Chen AM, Jahan TM, Jablons DM, Garcia J, Larson DA.Risk of cerebral metastases and neurological death after pathological complete response to neoadjuvant therapy for locally advanced nonsmall-cell lung cancer: clinical implications for the subsequent management of the brain.Cancer 2007;109:1668-75.34.Maas M, Nelemans PJ, Valentini V, et al.Long-term outcome in patients with a pathological complete response after chemoradiation for rectal cancer: a pooled analysis of individual patient data.Lancet Oncol 2010;11:835-44.35.Berger AC, Farma J, Scott WJ, et al.Complete response to neoadjuvant chemoradiotherapy in esophageal carcinoma is associated with significantly improved survival.J Clin Oncol 2005;23:4330-7.36.Chao YK, Chan SC, Liu YH, et al.Pretreatment T3-4 stage is an adverse prognostic factor in patients with esophageal squamous cell carcinoma who achieve pathological complete response following preoperative chemoradiotherapy.Ann Surg 2009;249:392-6.37. Donahue JM, Nichols FC, Li Z, et al.Complete pathologic response after neoadjuvant chemoradiotherapy for esophageal cancer is associated with enhanced survival.Ann Thorac Surg 2009;87:392-8.38. Park JW, Kim JH, Choi EK, et al.Prognosis of esophageal cancer patients with pathologic complete response after preoperative concurrent chemoradiotherapy.Int J Radiat Oncol Biol 2011;81:691-7.39.Reed VK, Krishnan S, Mansfield PF, et al.Incidence, natural history, and patterns of locoregional recurrence in gastric cancer patients treated with preoperative chemoradiotherapy.Int J Radiat Oncol Biol Phys 2008;71:741-7.
Current OnCOlOgy-VOlume 21, number 2, April 2014 Copyright © 2014 Multimed Inc.Following publication in Current Oncology, the full text of each article is available immediately and archived in PubMed Central (PMC).29E-mail:renatadalpino@gmail.com * BC Cancer Agency, Vancouver, BC.