Low toxic neoadjuvant cisplatin, 5-fluorouracil and folinic acid in locally advanced gastric cancer yields high R-0 resection rate

Abstract

Purpose

Neoadjuvant chemotherapy in locally advanced gastric cancer is effective, but is often associated with severe side effects, including fatal outcome. This study evaluates a combination of cisplatin, folinic acid and 5-fluorouracil (PLF) in terms of efficacy (R-0 resection rate) and toxicity.

Methods

Twenty-five patients with locally advanced gastric cancer who after extensive staging were deemed not suitable for curative resection underwent neoadjuvant chemotherapy. Three or four cycles of cisplatin (50 mg/m² days 1 and 15), folinic acid (200 mg/m² days 1, 8, 15 and 22), and 5-fluorouracil (2,000 mg/m² days 1, 8, 15 and 22) were administered. Cases with progressive disease were taken off the study. Two weeks after finishing chemotherapy resection was performed and all patients were enrolled in a structured follow-up.

Results

Of the patients, 22/25 finished chemotherapy and 20 of those underwent laparotomy. In 13/25 patients (52%) a R-0 resection and in three cases a R-1 resection were achieved. Four patients stayed irresectable. During 76 completed cycles of chemotherapy we observed five cases of WHO grade-III toxicity and no grade-IV toxicity.

Conclusions

The presented PLF protocol yields R-0 resection rates comparable to protocols like EAP (etoposide, adriamycin, platinum), but with a better safety profile allowing administration in an outpatient setting. Our study supports PLF as a reference neoadjuvant treatment for gastric cancer even outside of clinical studies.

Introduction

Gastric cancer remains a leading cause of cancer mortality despite a worldwide decline in incidence [Parkin et al. 2001]. In western countries the majority of at least 65% of the patients are still diagnosed in advanced UICC stages III and IV [Wanebo et al. 1996]. Only radical resection offers a chance of cure, but the results of surgery have reached a plateau of effectiveness. Even with extended surgery, including systematic lymphadenectomy, less than half of the patients with locally advanced gastric cancer achieve a macroscopic and microscopic tumor-free resection. In the German Gastric Cancer Study, which enrolled 1,999 patients, the R-0 resection rate was only 41.1% in UICC stage III [Roder et al. 1993]. R-0 resection, however, is the most important treatment-related prognostic factor.

Neoadjuvant chemotherapy aims at "downstaging" of the tumor in order to increase the chance for curative resection. Since the end of the 1980s there have been 18 phase-II studies testing this concept in locally advanced gastric cancer. Depending on staging—either by laparotomy or by clinical investigation—30–60% of patients became curatively resectable after neoadjuvant chemotherapy [Cascinu et al. 1998; Crookes et al. 1997; Fink et al. 1993; Furukawa et al. 1998; Gallardo-Rincon et al. 2000; Kelsen et al. 1996; Leichman et al. 1992; Melcher et al. 1996; Plukker et al. 1991; Schuhmacher et al. 2001; Schwartz et al. 1993; Wilke et al. 1989; Yonemura et al. 1993]. However, in several studies the applied regimens generated severe side effects, including chemotherapy-related death [Crookes et al. 1997; Kelsen et al. 1996; Leichman et al. 1992; Rougier et al. 1994]. Severe leuko- and thrombopenia was frequently observed using protocols containing adriamycin.

On the other hand, in metastastic disease a combination of weekly infusional high-dose 5-fluorouracil (5-FU), folinic acid and bi-weekly cisplatinum (PLF) yielded a response rate of over 50% with only mild toxicity [Wilke et al. 1996; Ychou et al. 1996]. Our aim was to test whether a PLF protocol in the neoadjuvant setting yields R-0 resection rates comparable to those of more toxic protocols like EAP (etoposide, adriamycin and platinum). Primary end points were the R-0 resection rate and the toxicity of the applied protocol. A secondary end point was the overall survival after neoadjuvant chemotherapy with and without resection.

Subjects and methods

The study group consisted of the Departments of Gastroenterology and Abdominal Surgery at the University Hospital of the Saarland, Homburg, as well as three other centers (the Department of Abdominal Surgery at the University Hospital of Magdeburg, the Theresien Hospital of Mannheim and the Municipal Hospital of Ludwigshafen). Between November 1997 and April 2001 all patients with gastric cancer or cancer of the gastric cardia in these four centers were screened for enrollment in the study. The study was approved by the local authorities and the ethics committee, the "Ethikkommission der Ärztekammer des Saarlandes" on 17 November 1997, and consequently by the according authorities of the collaborating centers. Written informed consent was obtained from each participant.

Inclusion criteria for screened patients

Inclusion criteria for screened patients included (1) primary cancer of the stomach or the gastric cardia of all histologic types, (2) not curatively resectable disease, i.e., UICC stages IIIB and IV M0 (for definition, see Table 1), (3) age between 18 and 70 years and (4) written informed consent.

Table 1.

Criteria to preoperatively define not curatively resectable disease

Either Infiltration of the tumor into the head of the pancreas, diaphragm, retroperitoneum, duodenum, celiac trunk or abdominal wall	And/or Pathologic lymph nodes in/at the ligamentum hepato-duodenale (=N3)
Pathologic Lymphnodes in Compartment 2 (=N2) together with a T3 or T4 tumor

Exclusion criteria

Exclusion criteria included (1) documented or suspected organ metastases, (2) presence of a secondary neoplasia, except resected basalioma or stage I cancer of the cervix, (3) recurrent cancer after gastrectomy, (4) severe co-morbidity (severe coronary heart disease, myocardial infarction during the past 12 months, cardiomyopathy, uncontrolled infection, systemic immune deficiency), (4) impaired hepatic function (serum-bilirubin >3 mg/dl and/or serum-albumin level <35 g/dl, (5) impaired renal function (serum creatinine >1.9 mg/dl and/or creatinine clearance <50 ml/min), (6) lack of compliance (e.g., drug or alcohol addiction), (7) expected survival time less than 3 months, (8) lack of physical performance (Karnovsky score <70%), (9) pregnancy, (10) carcinoma of the distal esophagus and (11) refusal of written consent.

TNM classification system

For this study the fourth edition of the TNM classification effective at the time of activation of the study was used [Hermanek and Sobin 1992] as well as the commentary on uniform use [Hermanek et al. 1993].

Diagnostic procedures and staging

All patients underwent an extensive staging including upper endoscopy, endosonography of the upper gastrointestinal tract, computed tomography of thorax and abdomen, sonography of the abdomen, bone scintigraphy and blood tests including parameters to check renal, hepatic and pulmonary function.

Irresectability in curative intention was defined according to the parameters shown in Table 1. All findings defining irresectability had to be confirmed by two independent staging procedures (e.g., endosonography and CT). Patients meeting these criteria were started on neoadjuvant chemotherapy after distant metastases had been ruled out.

In case of doubt—e.g., suspected infiltration of the diaphragm, diagnosed only by endosonography—surgical laparoscopy was performed. If the tumor infiltration was proven by this procedure, the patient underwent neoadjuvant chemotherapy; if not, the operation was performed without prior chemotherapy. In patients who during laparotomy turned out to have advanced disease contrary to initial staging, no resection was performed and neoadjuvant chemotherapy was started as soon as possible.

Chemotherapy

Each patient was treated via an intravenous port system with access to the subclavian vein. One chemotherapy cycle consisted of 28 days with the next cycle starting on day 36. Folinic acid was applied in a dosage of 200 mg/m² body surface via a 60 min infusion on days 1, 8, 15 and 22, followed by 5-fluorouracil 2,000 mg/m² as a continuous 24-h infusion on the same days.

On days 1 and 15 cisplatinum in a dosage of 50 mg/m² (dissolved in 1,000 ml NaCl 0.9%) was given as a 60-min infusion via a separate infusion system. For nephroprotection intravenous hyperhydratation was performed using 1,000 ml of NaCl 0.9% solution prior to cisplatinum and another 1,000 ml NaCl 0.9% solution thereafter over 3–4 h.

Antiemetic medication consisted of ondansetron plus dexamethasone and metoclopramide. After completion of two cycles re-staging was performed including upper endoscopy, CT and endosonography. If restaging revealed at least a "no change," therapy was continued for a third cycle. If after three cycles of neoadjuvant chemotherapy there was a documented further response, a fourth cycle could be added at the discretion of the responsible investigator. Surgery had to be performed 2 weeks after having finished the last chemotherapy cycle. Response to therapy was defined according to the WHO criteria for measurable diseases.

Toxicity

Toxic side effects were recorded according to the WHO classification system (WHO 1979). In case of major toxicity (>grade I) therapy adjustments were required according to algorithms fixed in the study protocol. For example, if there was mucositis or diarrhea of >grade 2 or leucopenia of grade 4 or thrombopenia of grade 3 or more, therapy was delayed until non-hematologic side effects subsided and/or the leucocyte count exceeded 2.5×10⁻⁹/l and the thrombocyte count exceeded 75×10⁻⁹/l.

Operation/surgical therapy

The resection of the tumor with a margin of at least 5 cm (=8 cm in situ) combined with systematic lymphadenectomy of compartments I and II and the resection of the omentum majus and minus was defined as standard. Compartment I had to be removed together with the stomach in an en-bloc-resection so that the pathologist was able to exactly define the pN stage. Only in case of an intestinal type and a distal location of the tumor was a subtotal resection allowed. In all other cases gastrectomy had to be performed. Splenectomy was mandatory in case of location in the upper half of the stomach or along the major curvature.

Statistical calculation

The statistical calculation was done with support of the Institute of Medical Biometrics, Epidemiology and Medical Informatics of the Saarland University (Prof. Dr. U. Feldmann). We postulated a 40% R-0 resection rate for the whole study population after neoadjuvant chemotherapy (based on an intention-to-treat analysis). Principally, all enrolled patients were non-resectable for cure according to the above-mentioned criteria. We defined a significance level of alpha=5% (P<0.05) and a presumed error of ß=10%, resulting in a statistical power of 90%. According to this the study population had to consist of at least 14 patients (one-sided chi-square test). The basis for statistical evaluation is an intention-to-treat analysis. The primary end points were the R-0 resection rate and toxicity. Survival is expressed as the median survival in months beginning with the day of entering the study.

Follow-up

Each patient was enrolled in a structured follow-up beginning on the day of surgery. Follow-up investigations were performed at months 3, 6, 12, 18, 24 and 36, including physical examination, laboratory tests and abdominal sonography. An abdominal CT and a chest X-ray had to be performed at months 6, 12 and 24, control upper endoscopy and bone scintigraphy were planned for months 12 and 24. Palliative chemotherapy for recurrent disease was not prohibited; the decision was made individually.

Results

Out of 127 patients with gastric cancer or cancer of the cardia between November 1997 and April 2001, 25 (20%) met the inclusion criteria. One hundred and two patients were ineligible due to having UICC stage IIIa or lower (n=41), metastatic disease (n=51), refusal to participate in a study (n=6) and withdrawal through the local surgeon (n=4).

The mean age of enrolled patients was 57 years (32–70). The male: female ratio was 2.1:1. Irresectability was determined by surgical staging in 14 patients and by clinical staging in the remaining 11 patients. In five cases surgical laparoscopy was performed, for instance to define an infiltration of the diaphragm by a cardia carcinoma. Twenty-two of 25 patients (88%) completed chemotherapy with an average of 3.3 cycles administered. Of those, 16 patients (73%) had a partial response, 4 (18%) achieved a stable disease, and one patient each had a minor response or progressive disease, respectively. Three patients did not complete chemotherapy due to progressive disease (1), death in a car accident (1) and supposed side effects of therapy (disturbance of vision, 1).

Resectability

Of the 22 patients having completed chemotherapy 20 underwent surgery, one patient refused the operation, having reached a partial response and feeling well, another one revealed rapidly progressive disease by the end of chemotherapy. In 13 of these 20 patients (65%) R-0 resection was achieved. Three patients turned out to have had R-1 resection and the remaining four were irresectable. Table 2 summarizes these results.

Table 2.

Outcome, maximum toxicity and survival of all recruited study patients. Hem hematological, Gastro gastrointestinal, neutrop neutropenia, thrombop thrombopenia, vomit vomiting, not applic not applicable, irresect irresectable, pts patients

Patient no.	Outcome	Maximum toxicity (WHO)	Type of maximum toxicity	Resection performed	Survival (all patients, in days)	Survival (R-0 resected patients, days)	Remarks
1	PR	2	Hem	R0	1222	1222
2	NC	1	Hem	No	223		Irresect
3	PR	2	Gastro, pulmonal	R1	561
4	PR	1	Hem	R0	343	343
5	PR	3	Hem (neutrop.)	R0	571	571
6	PR	1	Hem	R0	283	283
7	PR	2	Hem, gastro	No	440		Operation refused
8	NC	2	Hem	R1	320
9	PD	1	Hem, renal	Not applic	64		Death in PD
10	PR	1	Hem, gastro	R0	548	548
11	PR	2	Gastro	R0	401	401
12	PR	2	Hem	R0	1143	1143
13	PR	1	Hem, gastro.	R0	701	701
14	PD	1	Hem	No	198		PD after Ctx
15	PR	3	Hem (thrombop)	R0	768	768	Alive
16	PR	2	Hem	R0	480	480
17	NC	3	Gastro (vomit)	No	317		Irresect
18	MR	2	Renal	No	193		Irresect
19	NC	1	Hem	R0	969	969	Alive
20	PR	1	Hem, renal	R1	551
21	?	1	Gastro	Not applic	303		Death in accident
22	NC	1	Hem	R0	860	860	Alive
23	PR	2	Hem	No	480		Irresect
24	NC	3	Hem (neutrop)	Not applic	394		Therapy stopped
25	PR	3	Gastro (nausea)	R0	931	931	Alive
Median					480	701

Based on the whole study population, 13/25 (52%) were finally curatively resected (intention to treat analysis). Twelve of 20 patients who were operated on achieved a "downstaging" of at least one T or N category.

Toxicity

In general, during 76 completed cycles of chemotherapy there was no grade-IV toxicity observed. The predominant toxicity was of the hematological type, including four grade III events, followed by gastrointestinal toxicity with two grade III events.

Hematological toxicity

Grade 3–4 neutropenia occurred in two patients (8%). No febrile neutropenia was observed. Grade 3 anemia occurred in one patient, and grade 3 thrombocytopenia occurred also in one patient. In 15 patients grade II toxicity was noted, predominantly anemia and leukopenia (see Table 2).

Non-hematological toxicity

The most common non-hematological adverse events were (percent of the patients): nausea (20%), vomiting (12%) and diarrhea (8%). Grade 3 adverse events were nausea (4%) and diarrhea (4%) (see Table 2). In any case the toxic effects were reversible with no toxicity-induced termination of therapy. All other toxic side effects did not exceed grade II.

Survival

By 31 October 2002, after a median follow-up of 39 (18–58) months, 21 of the 25 patients had died. Overall median survival was 15.5 months. The remaining four patients (16%) are currently alive with a median overall survival of 28 months. Three patients are in complete remission; one patient developed a lymph node metastasis in the mesenterial root, which has been resected.

The median survival of those 13 patients who achieved a R-0 resection is 23 months, compared to a median survival of 10.5 months for those patients that failed to achieve R0 resection (R1 n= 3 and irresectable n=4), see Fig. 1. Survival of the three patients with R-1-resection was 10.5, 18 and 18.3 months, respectively.

Fig. 1.

Survival of operated patients (n=20). R-0 resected patients versus R-1 resected and irresectable ones (days). Note: median survival for 13 R-0 resected patients is 701 days (23 months) compared to 320 days (10.5 months) for the other seven patients who were operated on

Follow-up

Palliative chemotherapy was performed in three patients: in two patients because of local recurrence after 6 and 8 months and in the remaining one because of liver metastases after a disease-free interval of 11 months. During follow-up eleven patients died as a result of local recurrence; another ten died from metastatic disease.

Discussion

Our study is the first fully published one to demonstrate the efficacy of the combination of cis-platin, 5-fluorouracil and folinic acid in a cohort of locally advanced gastric cancer patients treated with neoadjuvant intention. More than half of the patients meeting defined criteria of irresectability were curatively resected after neoadjuvant chemotherapy. Furthermore, this neoadjuvant chemotherapy was well tolerated and not compromised by severe toxicity.

In phase-II-studies the PLF protocol had yielded response rates of 66 and 52% in metastatic disease with acceptably low toxicity [Wilke et al. 1996; Ychou et al. 1996]. In a randomized multicenter trial the combination of infusional 5-FU and cisplatinum proved to be equivalent to sequential high-dose methotrexate, 5-FU and doxorubicin (FAMtx) as well as to etoposide, leucovorin, and 5-FU (ELF), representing former "standard therapies" in metastastic disease [Vanhoefer et al. 2000].

Neoadjuvant treatment has gained growing attention since the end of the 1980s, driven by the fact that curative resection of the tumor is the single most important prerequisite for long-term survival or cure. Neoadjuvant chemotherapy aims at "downstaging" of the primary tumor to facilitate complete resection and at elimination of lymph node metastases.

Additionally, systemic chemotherapy might eliminate or at least reduce the spread of tumor cells prior to an operation. Recently, Walsh et al. demonstrated a reduction of cytokeratin-18 positive tumor cells in the bone marrow from 66 to 36% in patients treated with neoadjuvant chemotherapy [Ryan et al. 2002].

Since 1989 there have been 18 published studies on neoadjuvant chemotherapy, which may be divided in those including potentially resectable patients (in the curative sense) or those including only irresectable patients. Further, the individual procedures to define irresectability differ.

R-0 resection

In six studies including a total of 172 definitely irresectable patients, most of them defined by laparotomy [Cascinu et al. 1998; Gallardo-Rincon et al. 2000; Lerner et al. 1992; Plukker et al. 1991; Wilke et al. 1989; Wils et al. 1991], R-0 resections were achieved in 8–44%.

Combinations used for neoadjuvant chemotherapy were etoposide, adriamycin and platinum (EAP) (n=2), a combination of the ELF protocol with platinum (n=2), methotrexate combined with 5-FU (n=1) and FAMtx (n=1).

Another six studies included altogether 208 irresectable and potentially resectable patients [Fink et al. 1995; Kelsen et al. 1996; Melcher et al. 1996; Schuhmacher et al. 2001; Siewert et al. 1998; Yonemura et al. 1993], with 10 to 56 patients per study, mostly of clinical stages IIIA, IIIB and IV, whenever specified. In these studies the R-0 resection rates varied between 41 and 80%.

The PLF protocol used in the study of Siewert et al. [Siewert et al. 1998] differed from ours regarding the length of a single cycle (6 weeks vs. 4 weeks in our trial), but resulted in similar cumulative doses. Staging for inclusion relied on EUS in this study: Cases of eus-T3 or -T4 stage patients were randomised for either surgery alone or neoadjuvant therapy followed by operation, so that a substantial part of potentially resectable patients in a T3-N1 (=UICC IIIA) stage were treated with chemotherapy.

The remaining six studies administered neoadjuvant chemotherapy to potentially resectable patients [Crookes et al. 1997; Leichman et al. 1992; Lowy et al. 1999; Rougier et al. 1994], covering a total of 210 (27–59) patients. As to be expected, R-0 resection rates (60–76%) were consistently higher than in the other trials.

So far, a benefit of neoadjuvant therapy in potentially resectable patients with gastric cancer has not been demonstrated—data of a large multicenter trial ( the MAGIC study) dealing with this issue will be published in 2003.

In our study the focus was on patients not deemed resectable for cure in order to increase resectability. Irresectability was either defined by laparotomy or by extensive clinical staging including endosonography, CT and, in case of conflicting results, surgical laparoscopy. The investigator had to prove either infiltration of defined neighboring organs or the presence of distant pathologic lymph nodes in the D2 or D3 compartment together with a locally advanced tumor with two independent staging investigations in order to classify a patient as irresectable. This procedure reduces the need for laparotomy without compromising the accuracy of the preoperative staging.

With a 65% R-0 resection rate (per protocol analysis) and a 52% R-0 resection rate (intention to treat analysis), the applied PLF protocol is at least as effective as the neoadjuvant protocols described in the literature.

Toxicity

In general, the studied neoadjuvant chemotherapy, although comparable in efficacy, differed substantially in toxicity. Two studies [Crookes et al. 1997; Leichman et al. 1992] combined preoperative systemic with postoperative intraperitoneal chemotherapy, which was probably responsible for a high rate of major toxicity events (grade III and IV) and four therapy-related deaths among 115 treated patients (3.5%).

In two studies restricted to preoperative therapy two cases of therapy-related deaths are reported in 86 patients (2.3%) [Kelsen et al. 1996; Rougier et al. 1994].

Especially the EAP protocol, which was the first to demonstrate effective "downstaging," is burdened with a predominant hematological toxicity (30–43% grade III and 18% grade IV) [Fink et al. 1995; Wilke et al. 1989].

Our PLF protocol generated a surprisingly low toxicity rate with grade-III events in only 20% of the patients and no grade-IV toxicity. All observed major toxic effects occurred in cycles two and three.

Except for a single patient, who was treated for 2 days with intravenous infusions because of diarrhea and vomiting, there was no need for inpatient treatment for side effects of the chemotherapy, indicating that this protocol is suitable for an out-patient setting.

Siewert et al. [Siewert et al. 1998] report a similarly low toxicity rate—18% grade III and 2% grade IV—in their study on 49 patients using a similar PLF protocol. Whenever details regarding toxicity were given in the remaining studies mentioned above, toxicity rates were markedly higher than in our study.

Survival

Median survival of untreated patients with locally advanced gastric cancer without evidence of metastases (stages UICC III B and IV M0) is about 6 to 9 months.

Focusing on definitely irresectable patients—staged by laparotomy—Wilke et al. [1989] observed a median survival of 18 months for all patients and 24 months for disease-free individuals after neoadjuvant chemotherapy. Other authors reported a median survival of 13.3 months for patients with initial irresectable cancer who responded to neoadjuvant chemotherapy [Gallardo-Rincon et al. 2000] and of 22 months in those patients that became resectable [Plukker et al. 1991].

In a group of patients of UICC stages IIIA, IIIB and IV M0, Schumacher et al. demonstrated a median survival of 19.1 months for the whole study population and of 28.4 months for patients in whom a R-0 resection was achieved after neoadjuvant chemotherapy [Schuhmacher et al. 2001].

In this context our median survival rates of 15.5 months for the whole study population and of 23 months for R-0 resected patients are comparable to those reported in the literature and underline the efficacy of the PLF protocol. Furthermore, our data show that the reduction in toxicity does not compromise efficacy.

In conclusion, based on the moderate toxicity and good efficacy, our study supports the PLF protocol as a reference neoadjuvant treatment in locally advanced gastric cancer. Compared to former studies our results are based on a more accurate clinical definition of irresectability. In more than half of the patients treated with PLF in this way, R-0 resection could be performed and toxicity was mild. The moderate toxicity profile of the PLF protocol allows the treatment of the majority of patients on an outpatient basis.