For locoregional disease, surgery remains the mainstay of treatment. Various reviews have reported 5-year overall survival (OS) rates from 10% to 40% with surgical resection alone.^78,120 Primary radiation therapy was previously used for local tumor control, but—in one large series—the 3-year survival after radiotherapy alone was only 6%.⁴⁸ For metastatic disease, chemotherapy alone has moderate activity but is not curative.

Given the activity of all three modalities, numerous studies have combined them in distinct preoperative strategies, also called neoadjuvant therapy. Multimodality approaches have included chemotherapy or radiation or concurrent chemoradiotherapy followed by surgery in an effort to improve the dismal prognosis of this aggressive cancer.

The results of these studies have been mixed and their combined outcomes have failed to elevate any of these preoperative strategies to a clear standard for resectable esophageal cancer. However, trials involving preoperative chemotherapy and chemoradiotherapy have demonstrated a trend toward improved survival over surgery alone. Based on these data, many clinicians now treat locoregional disease with preoperative multimodality therapy.

Although several chemotherapeutic agents have modest antitumor activity in esophageal cancer, the duration of response to both single agents and combination regimens is only generally 4 to 6 months. In the palliative setting, this approach may be considered appropriate treatment. However, chemotherapy is rarely used alone with curative intent.

The activity of single-agent drugs in esophageal carcinoma is summarized in Table 159-1. Commonly used drugs include cisplatin, 5-fluorouracil (5-FU), and mitomycin, with single-agent response rates ranging from 10% to 25%. Newer agents include the oral 5-FU prodrugs (capecitabine and S-1), the taxanes (paclitaxel and docetaxel), the new platinum compound oxaliplatin, and irinotecan, with response rates of 15% to 45%. Some of these trials have primarily enrolled patients with gastric cancer, but they also include patients with cancer of the gastroesophageal (GE) junction. Despite the variable response rates reported for some single agents, the confidence limits largely overlap across trials in most cases.

Since its introduction in 1980, cisplatin has become the foundation for most combination chemotherapy regimens for esophageal cancer. In contrast, carboplatin has shown a disappointingly low 0% to 9% response rate in both squamous cell carcinoma^114,138,157 and adenocarcinoma.⁴⁹ The newer platinum analog oxaliplatin has been evaluated in recent combination chemotherapy trials but not as a single agent in gastric or esophageal cancer.

Several oral prodrugs of 5-FU have undergone evaluation in patients with advanced gastric cancer and cancer of the GE junction. One such drug is capecitabine, which produces response rates of 19% to 34%.^77,100 Another oral prodrug, S-1, has also recently undergone investigation. S-1 is an oral formulation of tegafur (a 5-FU prodrug), 5-chloro-2,4-dihydroxypyridine (a dihydropyrimidine dehydrogenase inhibitor, which prevents the degradation of 5-FU), and potassium oxonate (which is thought to reduce the diarrhea associated with 5-FU therapy). As a single agent, it produces response rates of 28% to 49%.^{26,34,123,146}

Boku and associates recently reported a randomized study in patients with advanced gastric cancer of infusional 5-FU versus S-1 (a third arm consisted of cisplatin/irinotecan).²⁶ S-1 was
shown to be noninferior to infusional 5-FU in terms of response rate and survival.

Paclitaxel has been administered in various schedules, including infusions of 96, 24, and 3 hours as well as 1 hour. The initial phase II trial of 24-hour paclitaxel showed an overall response rate of 32%.⁷ A subsequent evaluation of a weekly 1-hour infusion of paclitaxel produced a response rate of 15%, with significantly less hematologic toxicity than has been seen with more protracted infusions.⁸⁵ While there are no studies comparing paclitaxel and docetaxel directly, three phase II evaluations of single-agent docetaxel report response rates of 17% to 20%.^50,72,121

Given the modest activity of single agents in esophageal cancer, combination chemotherapy has been extensively studied (Table 159-2). In metastatic disease, cisplatin-containing regimens have shown 25% to 35% activity in squamous carcinoma. In locoregional disease, response rates as high as 45% to 75% have been reported. Disappointingly, these responses have been no more durable than those of single agents.

The combination of cisplatin/infusional 5-FU has been studied extensively, with toxicity consisting mainly of mucositis, diarrhea, and myelosuppression. Despite the common use of this regimen, only a single phase II trial has compared cisplatin with the combination regimen in patients with advanced squamous cell carcinoma.²⁴ While there was a higher response rate in favor of the combination regimen, there was no statistically significant survival benefit. In addition, the combination group had a significantly higher rate of treatment-related deaths (16% versus 0%).

Randomized trials have compared numerous other drug combinations to cisplatin/5-FU. For example, the ECF regimen (epirubicin/cisplatin/infusional 5-FU) has been compared with the FAMTX regimen (bolus 5-FU/doxorubicin/methotrexate).¹⁷⁴ The ECF arm achieved superior median survival (8.9 months versus 5.7 months), response rate (45% versus 21%), and quality of life at 24 weeks compared with FAMTX.

A subsequent study compared the ECF and MCF regimens, where mitomycin was substituted for epirubicin in previously untreated patients with advanced esophageal cancer.¹⁴² The study reported no significant differences in response rate or median survival but found that quality of life was better maintained with ECF. The equivalent response rates and outcomes that were seen with both of these regimens have led some to question whether either regimen offers any advantage over the conventional two-drug standard. The data from the ECF trials, however, support the use of continuous infusion 5-FU over bolus 5-FU as well as the lower doses of cisplatin more commonly used in the two drug 5-FU/cisplatin combination.

In the recent REAL-2 study, Cunningham and colleagues from the United Kingdom compared the ECF regimen with the ECX regimen (which involves the substitution of 5-FU with capecitabine), the EOF regimen (substitution of oxaliplatin for cisplatin), and the EOX regimen (a double substitution of both capecitabine and oxaliplatin) in patients with locally advanced esophageal, GE junction, or gastric cancer.⁴⁰ All the combinations had similar response rates and toxicities and the EOX regimen was associated with improved median OS compared to the ECF regimen (11.2 versus 9.9 months), leading the investigators to propose that the EOX regimen could replace the ECF regimen in future trials.

The oral 5-FU prodrug S-1 has also been evaluated in combination regimens. A phase II evaluation of S-1/cisplatin reported a response rate of 51%, with a median OS of 10.9 months.⁹ A randomized phase III Japanese trial in advanced gastric cancer of S-1 versus S-1/cisplatin was also recently reported in abstract form.¹²³ The combination regimen was associated with a higher response rate (54% versus 31%) and median OS (13.2 months versus 11.2 months). Based on their data, the authors proposed that S-1/cisplatin should become a reference regimen for advanced gastric cancer.

It should be noted that—because of genetic polymorphisms in hepatic cytochrome enzymes responsible for chemotherapy metabolism—the maximum tolerated dose of S-1 is smaller in non-Japanese than in Japanese patients.⁷⁶ Therefore, the results of this Japanese trial await confirmation in a recently completed global phase III evaluation of S-1/cisplatin versus 5-FU/cisplatin in advanced gastric cancer patients that excludes Japanese patients.

Other investigators have combined cisplatin with paclitaxel, both with and without 5-FU.^{81,83,133,134} Response rates ranged from 43% to 50%, but toxicity included significant diarrhea, neurotoxicity, and myelosuppression.

The addition of docetaxel to cisplatin/5-FU has also been evaluated. A recent randomized trial in gastric cancer and cancer of the GE junction compared the DCF regimen (docetaxel/cisplatin/infusional 5-FU) to cisplatin/infusional 5-FU.¹⁶⁹ Although response rate and time to progression (TTP) were improved with the three-drug regimen over the two-drug regimen, median OS was only slightly improved (9.2 versus 8.6 months). In addition, the three-drug regimen was associated with significantly more toxicity, including a grade 3/4 neutropenia rate of 82% (versus 57%) and febrile neutropenia in 29% of patients (versus 12%).

A slight variant of the DCF regimen used by Van Cutsem et al.¹⁴³ (termed TCF, employing a 14- versus 5-day 5-FU infusion) was also recently compared to the ECF regimen in a phase II randomized trial. TCF appeared to result in a superior response rate (the primary endpoint) compared with ECF but the toxicity, particularly rates of neutropenia and neutropenic fever, was again substantial. This phase II trial also included a third arm with the TC (docetaxel/cisplatin) regimen. Activity and survival were comparable between the TC and TCF arms and there was a suggestion of superior toxicity profile for the two-drug TC regimen compared with the three-drug TCF regimen. Similarly, another randomized phase II study demonstrated comparable activity for a regimen of DF (docetaxel/5-FU) versus ECF, although the study was not powered for a head-to-head comparison of both regimens.¹⁶⁴

Irinotecan is another active agent that has been combined with mitomycin, 5-FU/leucovorin, or cisplatin, with response rates ranging from 30% to 65%.^{8,67,82,137,151} In a randomized phase II trial, Pozzo and colleagues¹³⁷ compared a regimen of FUFIRI (weekly irinotecan/5-FU/leucovorin) with irinotecan/cisplatin administered every 3 weeks in patients with advanced gastric cancer and cancer of the GE junction. The FUFIRI regimen was associated with a superior response rate, TTP, and OS and less neutropenia than the irinotecan/cisplatin arm. This led to a subsequent phase III trial of FUFIRI versus cisplatin/5-FU. Both regimens had comparable efficacy but there was less neutropenic fever and grade 3/4 stomatitis and nausea in the FUFIRI arm.⁴² Only the incidence of grade 3/4 diarrhea was increased in the FUFIRI arm, although more patients withdrew from the cisplatin/5-FU arm than the FUFIRI arm (22% versus 10%, p = 0.004) for drug-related adverse events.