Adjuvant treatment

Abstract

Exocrine pancreatic cancer (pancreatic ductal adenocarcinoma) is one of the leading causes of cancer deaths in the western world, accounting for 5% of all cancer-related deaths. Only a small percentage of patients with pancreatic cancer are able to undergo potentially curative resection, even in specialized centres, and prognosis remains poor after successful surgery. Over the last few years efforts have been directed towards the development of adjuvant therapies in attempts to improve outcome. The main trials of adjuvant chemotherapy, chemoradiotherapy and chemoradiotherapy with follow-on chemotherapy are described in this paper, followed by the results of the ESPAC-1 trial and the status of ESPAC-2 and -3 trials.

Introduction

Exocrine pancreatic cancer (pancreatic ductal adenocarcinoma) is one of the leading causes of cancer deaths in the western world, accounting for 5% of all cancer-related deaths ¹. Worldwide there are 227 000 deaths per year. Only 10–15% of patients with pancreatic cancer are able to undergo potentially curative resection, even in specialized centres ². Following successful surgery, prognosis remains poor, as the median survival is around 13–15 months and the 5-year survival rate is 15–20% ³,⁴. Attempts at more radical pancreatic resections and extended lymphadenectomy, although feasible without excessive morbidity and mortality, have failed to produce any convincing improvement in survival ⁵. This may be due, in part, to the fact that the majority of disease recurrences occur locally or in the liver ⁶,⁷,⁸. Over the last few years, therefore, efforts have been directed towards the development of adjuvant therapies in an attempt to improve outcome. The main trials of adjuvant chemotherapy, chemoradiotherapy and chemoradiotherapy with follow-on chemotherapy will be described, followed by the results of the ESPAC-1 trial and the status of ESPAC-2 and -3 trials.

Systemic chemotherapy

Until the European Study Group for Pancreatic Cancer (ESPAC) 1 trial ⁹,¹⁰, there were few adequately powered, randomized studies assessing adjuvant therapy in pancreatic cancer (Table I). This multicentric, randomized, two-by-two factorial design trial accrued 541 patients between February 1994 and June 2000 from 61 cancer centres in 11 countries. Besides the two-by-two factorial design allocation (i.e. observation, chemoradiotherapy alone, chemotherapy alone and both), randomization outside of the two-by-two factorial design, into one of the main treatment comparisons (i.e. chemotherapy versus no chemotherapy and chemoradiotherapy versus no chemoradiotherapy) was permitted.

Table I. Adjuvant systemic chemotherapy: randomized controlled trials.

Series	Period	n	Regimen	Median survival (months)	Actuarial survival (%) 1 year	Actuarial survival (%) 2 year	Actuarial survival (%) 3 year	Actuarial survival (%) 5 years
Bakkevold et al. [12]	1984–1987	61	5-FU/DOX//MMC	23	70		27	4
		31	–	11 (p=0.02)	45		30	8
ESPAC 1-interim [9]	1994–2000	238	5-FU	19.7
		253		14 (p 0.005)
ESPAC-1 Final [10]	1994–2000	147	5-FU	20.1		40		21
		142		15.5 (p=0.009)		30		8
Takada et al. [13] (pancreas)	1986–1992	81	MMC/5-FU					11.5
		77	–					18 (p=NS)
Neuhaus et al. [15]	1998–2004	179	Gemcitabine	Results awaited	Awaited	Awaited	Awaited
		177	–

5-FU, 5 fluorouracil; DOX, doxorubicin; MMC, mitomycin C.

Chemotherapy was administered using 5-fluorouracil (5-FU) given intravenously at a dose of 425 mg/m² daily for 5 days, over a period of 6 months. In the interim analysis of 541 patients after a median follow-up of 10 months ⁹, there was evidence of survival benefit for adjuvant chemotherapy. The median survival was 19.7 months in the 238 patients with chemotherapy versus 14 months in the 235 patients who did not receive chemotherapy (p=0.0005).

Influence of resection margin on survival was assessed, with 101/541 (19%) having positive microscopic resection margins (R1) ⁵. The median survival was 10.9 months in those who had R1 resection versus 16.9 months for patients with R0 resection, establishing resection margin status as an influential prognostic factor. Irrespective of the R0/R1 status, there was survival benefit for chemotherapy, with R0 patients having longer survival following chemotherapy in comparison with the group that did not receive chemotherapy. This difference was less apparent for the smaller subgroup of R1 patients, despite there being no significant heterogeneity between the two, i.e. R0 and R1 groups.

The final analysis of the ESPAC-1 trial assessed the 289 patients randomized using the two-by-two factorial design, and followed up for a median of 47 months ¹⁰. The conclusions were similar, with the median survival being 20.1 months among the 147 patients who were assigned to the chemotherapy group and 15.5 months among the 142 patients who did not receive chemotherapy (p=0.009) (Figure 1). The 2-year and 5-year survival estimates were 40% and 21%, respectively, in the chemotherapy group and 30% and 8% in the group without chemotherapy. A significant survival benefit for chemotherapy was demonstrated, irrespective of whether the patients were randomized to receive chemoradiotherapy or not. Prognostic factors that had significant adverse influence on survival were the differentiation of tumours (p<0.001), lymph nodal involvement (p<0.001), and a maximum tumour size of >2 cm (p=0.003). Prognostic factors of borderline significance were current smoking (p=0.07), R1 status (p=0.10), and presence of involved adjacent structures on histologic analysis (p=0.10). The median time to recurrence was 15.3 months in the chemotherapy group versus 9.4 months in patients who did not receive chemotherapy (p=0.02).

Figure 1. .

Kaplan–Meir estimates of survival according to whether or not patients received chemoradiotherapy (A) or chemotherapy (B) in the ESPAC-1 trial final results. CRT, chemoradiotherapy; CT, chemotherapy.

The influence of the type of surgery and the presence of complications on survival (in conjunction with clinicopathological variables) was investigated using the Cox proportional hazard model ¹¹. The final model was based on 481 ESPAC-1 patients who had undergone either Kausch-Whipple (KW) or pylorus-preserving KW (PPKW). Neither the type of surgery nor the presence/absence of complications significantly influenced survival. Benefit for chemotherapy was demonstrated, irrespective of these two factors. Based on these findings, it is recommended that the occurrence of postoperative complications should not deter patients from participating in adjuvant treatment trials.

Bakkevold et al. ¹² from Norway conducted a much smaller randomized controlled trial (RCT) between 1984 and 1987, randomizing 61 radically resected patients to receive either postoperative adjuvant combination chemotherapy or no chemotherapy, i.e. control group. Of the 61 patients, 47 had pancreas cancer while the rest had periampullary cancer. The chemotherapy regimen employed comprised 5-FU (500 mg/m²), doxorubicin (40 mg/m²) and mitomycin C (6 mg/m²) given every 3 weeks for a total of six cycles. Chemotherapy was associated with an increased median survival (23 months versus 11 months in the control group; p=0.04) and delay in the incidence of recurrence in the first 2 years after radical surgery, but an increased cure rate was not observed. Difficulties with this regimen included substantial toxicity: of the 30 patients randomized to receive chemotherapy only 13 managed to complete the six courses. Moreover, after the first course, 16 of 22 patients needed hospitalization for toxic side effects. Also, as the study analysis (survival) pooled patients with pancreas and periampullary cancer, it is difficult to draw conclusions on the usefulness of adjuvant chemotherapy solely for pancreas cancer.

Another RCT from Japan by Takada et al. ¹³ shares the drawbacks of the Norwegian trial, as this study also enrolled patients with pancreatic, gall bladder, bile duct and ampulla of Vater cancers. However, analysis was performed for each specific disease. Between April 1986 and June 1992, 508 patients were enrolled, including 173 resected pancreatic cancer patients. Patients were assigned to receive either combination chemotherapy (mitomycin C and 5-FU) or to the control group. The chemotherapy group received mitomycin C (6 mg/m² i.v.) at the time of surgery and 5-FU (310 mg/m² i.v.) in two courses for 5 consecutive days during postoperative weeks 1 and 3, followed by 5-FU (100 mg/m² orally) daily from postoperative week 5 until disease recurrence. After ineligible patients were excluded, data from 158 patients with pancreatic cancer was analysed (58 in the chemotherapy group and 60 in the control group). The 5-year survival rate was 11.5% in the chemotherapy group and 18.0% in the control group, and there was no statistically significant difference in survival between the two groups. Likewise, there was no significant difference in the disease-free survival time or time to recurrence. The equivocal results between the chemotherapy and control group could have been due to the use of oral 5-FU, which because of its hepatic metabolism has very poor efficacy compared with intravenously administered 5-FU or specially designed oral fluoropyrimidines ¹⁴.

In a recent RCT (July 1998 to December 2004), Neuhaus et al. ¹⁵ randomized 368 patients to receive either adjuvant chemotherapy with gemcitabine or observation following resection of pancreatic cancer. Preliminary results have revealed improved survival in the gemcitabine arm (14.2 months) compared with the observation arm (7.5 months; p<0.05). Final results of this trial are expected in late 2005.

Kurosaki and Hatekayama ¹⁶, in a prospective trial (1998–2003), enrolled 21 patients with node-positive pancreatic cancer following curative resection into a gemicitabine monotherapy adjuvant trial. Nine patients received chemotherapy with biweekly gemcitabine (1000 mg/m²), while the remaining 12 patients only underwent resection. Chemotherapy was well tolerated. The overall 1- and 2-year cumulative survival rates of the chemotherapy group were 86% and 50%, respectively, while those of the surgery only group were 75% and 0%, respectively. The median survival of the chemotherapy group was 20.3 months versus 15.4 months in the surgery only group (p=0.0084). The disease-free interval was also significantly greater in the chemotherapy group compared with the surgery-alone group (p=0.0244). Splinter et al. ¹⁷, in the 1980s, examined the use of 5-FU, doxorubicin and mitomycin C (FAM) in 16 patients and compared them with a historical control group of 36 patients. The 3-year survival rate was 24% for the FAM group versus 28% in the control group, implying no benefit from adjuvant chemotherapy.

Baumel et al. ¹⁸ reported a retrospective series involving 787 patients who had undergone pancreatic resection for cancer. In all, 43 patients were given postoperative chemotherapy, although there was no standardization of the chemotherapy regimens used. The median survival time was 12.4 months for patients in the surgery only arm and 11.5 months for the adjuvant chemotherapy arm.

Of the few trials looking at the role of adjuvant chemotherapy, the ESPAC-1 trial is the largest and not confounded by the inclusion of periampullary/ biliary neoplasms. The findings of this trial are highly significant and will direct the focus of future studies to the use of chemotherapeutic agents alone. To address this issue the ESPAC-3 trial is recruiting patients and has randomized 122 patients so far to receive either 5-FU or gemcitabine post surgery.

Regional chemotherapy

To reduce the side effects of systemic chemotherapy, regional delivery of chemotherapeutic agents to the target site has been carried out (Table II). In one of the earliest series, Ishikawa et al. ¹⁹ directed chemotherapy to the liver, as this is a common site of treatment failure following curative resection. Both the hepatic artery and portal vein were cannulated after extended pancreatectomy and 5-FU was given by continous infusion via these indwelling catheters. Of the 20 patients who were treated thus, the 3-year survival rate was 54%, and the cumulative rate of death from hepatic metastasis was 8%. The drawback of this study was that the results were compared to historical controls, against whom the regional treatment regimen was superior. In a later publication by the same group ²⁰, a comparison was made between 27 patients who received liver perfusion chemotherapy against 67 patients who did not. All patients had undergone extended pancreatectomy between 1987 and 1995. The cumulative survival rate was significantly better in the regional chemotherapy group at 1 year (92% in the chemotherapy group versus 62% in the surgery arm; p<0.05) and 3 years (51% vs 35%; p<0.05) but there was no difference at 5 years (41% vs 25%).

Table II. Adjuvant regional therapy for pancreatic cancer.

Series	Year	n	Regimen	Median survival (months)	Actuarial survival (%) 1 year	Actuarial survival (%) 3 year	Actuarial survival (%) 5 year
Ishikawa et al. 19	1994	20	HAI + PVI (5-FU)	–	–	54	–
Ozaki et al. 21	1994	24	IORT + HAI +/− PVI (MMC)	–	–	–	32
Link et al. 24,24	1997	20	CAI	21	–	–	–
		29	–	9.3	–	–	–
Ishikawa et al. 20	1997	27	HAI + PVI (5-FU)		92	51	41
		67			62 (p<0.05)	35 (p<0.05)	25 (p=ns)
Beger et al. 26	1999	24	CAI(MX/5-FU/CP)	23	–	–	54 (4 year; R0 resection)
		25	–	10.5 (p<0.001)			9.5 (4 year; R0 resection)
Ozaki et al. 22	2000	27			31.1		31%
Papachristou et al. 25	2003	31	CAI (MX/5-FU/CP)	21

HAI, hepatic arterial infusion; PVI, portal vein infusion; CAI, coeliac artery infusion; IORT, intraoperative radiotherapy; 5-FU, 5-fluorouracil; MX, mitoxantrone; CP, cisplatin.

In 1994, Ozaki et al. ²¹ reported a 5-year survival of 32% in 24 patients who had resection and who were then further treated with intraoperative radiotherapy (IORT) as well as regional chemotherapy. In a later publication ²² in 2000, they published their results for 30 patients (November 1983 to January 1993) treated by extended pancreatectomy, IORT and hepatic artery or portal vein infusion of mitomycin C followed by systemic bolus administration. The 5-year survival rate for 27 patients (excluding 3 with metastases to the liver, peritoneum, or lung) was 31%, with a median survival of 31.1 months. In a further analysis of 19 patients with regional nodal metastases, the 1-year survival rate was 95%, 3-year survival rate 50% and 5-year survival rate 28%.

Link et al. ²³,²⁴ administered coeliac artery infusion chemotherapy to 20 patients, 18 with pancreatic ductal adenocarcinoma and 2 with cystadenocarcinoma. The agents used were mitoxantrone, 5-FU plus folinic acid and cisplatinum given for 6 cycles. The median survival time of the treated group was 21 months, which was significantly better (p<0.0006) compared to 9.3 months in a group of matched historical controls. Only 15% of patients in the treated group developed liver metastases, prompting the authors to conclude that coeliac artery infusion chemotherapy both improves survival and decreases/delays the occurrence of liver metastases. Papachristou et al. ²⁵ likewise administered coeliac artery infusion chemotherapy to 31 patients, and found a median survival of 21 months in the treated group versus 9.3 months in historical controls (p<0.0003). The sites of treatment failure were pancreas bed (50%), intraperitoneal (40%) and liver (15%).

Beger et al. ²⁶ compared results between 26 patients (December 1992 to July 1997) treated with coeliac artery infusion (CAI) chemotherapy (mitoxantrone, folinic acid, 5-FU and cisplatin) after resection and 25 patients (January 1993 to May 1996) who received no adjuvant chemotherapy after resection. Systemic side effects were minimal, with 8% WHO grade III side effects (mainly gastrointestinal ulceration) and no grade IV toxicity. Excluding two patients with tail lesions, the remaining 24 patients with head lesions who received CAI had a median survival of 23 months. This was significantly better than the 10.5 months survival noted in the surgery only group (p<0.001). When patients who underwent R0 resection were analysed, the median survival in the surgery only arm was 12 months, whereas in the CAI group the median survival time was not reached after an observation period of 48 months (p<0.0001). In the CAI group, 18 patients developed recurrence of tumour, with 17% suffering recurrence in the liver and 83% experiencing local recurrence. Thus, although CAI provided hepatic protection, its effectiveness in local control was questionable.

Yamaue et al. ²⁷ looked into the effectiveness of hepatic artery infusional chemotherapy in preventing liver metastases in the subset of patients with portal vein invasion. Following radical surgery, this group has an enhanced risk of developing liver metastases ²⁸,²⁹,³⁰. The chemotherapeutic agent used was determined by assessing the chemosensivity of fresh human pancreatic cancer cells and the drugs tested were cisplatin, mitomycin C, adriamycin, 5-FU and etoposide. Nine patients received chemotherapy while five did not. None of the patients in the chemotherapy group developed liver metastases, in contrast to the no chemotherapy group, wherein all five patients developed liver metastases. The sites of failure in the chemotherapy group were locally and in the peritoneum. The mean survival of patients in the chemotherapy group was significantly longer (25.6 months) than the control group (9.3 months; p<0.05). The problems with this study are the small numbers and the fact that there was no uniformity in the agents used.

Results from trials on regional chemotherapy appear promising in terms of reducing liver metastases and improving survival, but there are no randomized controlled trials on this subject. The ESPAC-2 trial is a multicentre, prospective randomized controlled phase III trial to compare adjuvant postoperative intra-arterial chemotherapy and radiotherapy (arm A) with surgery alone (arm B) in resectable pancreatic ductal adenocarcinoma and advanced periampullary cancer. Patients in arm A will receive a total of six cycles of coeliac artery infusion and a total of 54 Gy external beam radiation therapy (EBRT). The study has recruited 70 patients so far and aims to recruit 100 patients overall. It should provide a definitive answer to the use of regional therapy for pancreatic cancer in the adjuvant setting.

Chemoradiotherapy

Radiation treatment, either by external beam radiation or intraoperatively, has been given with the idea of controlling any microscopic residual disease, as most recurrences following pancreaticoduodenectomy occur at the site of resection. Radiation has been given preoperatively, intraoperatively or postoperatively and often with concurrent chemotherapy both for radiosensitization and to address systemic micrometastases.

Intraoperative radiotherapy

The irradiation of the upper abdomen by EBRT causes considerable toxicity and IORT can reduce this, sparing normal tissues. The surrounding tissues can either be displaced or shielded, thereby allowing the delivery of larger radiotherapy doses in a single fraction to volumes harbouring tumour cells.

Fossati et al. ³¹, in a retrospective study (January 1985 to September 1992), reviewed data from 54 patients with both resectable and unresectable disease who received IORT and compared them to a control group of resected patients who did not receive IORT. Although there was significantly better local control achieved in resected patients who received IORT (25% in the IORT group versus 55.8% in the non-IORT group), this did not translate into improved survival. Similar results were found by Hiraoka et al. ³² in 1990, i.e. no increase in survival, when they compared 15 patients who received IORT after standard resection against 19 who did not receive IORT. Subsequently, they performed extended operation, followed by IORT (from the diaphragm above to the inferior mesenteric artery below) in 14 patients. On follow-up, no radiation-related complication occurred and the 5-year cumulative survival was 33.3%.

Between January 1986 and April 1995, Coquard et al. ³³ treated 25 patients with IORT (dose range 12–25 Gy), followed by EBRT (44 Gy) in 20 patients. Concurrent chemotherapy on days 1–4 with 5-FU was given to seven patients. The overall survival at 1 year was 56%, at 2 years was 20% and at 5 years was 10%. Nine patients developed locoregional relapse while 14 had metastatic disease. The small number treated, retrospective nature and the lack of a control/comparison group are the drawbacks of this study.

Zerbi et al. ³⁴ (Table III) compared IORT to surgery only. The IORT group had a significantly improved median time to relapse (13 vs 8 months), better local control (73% vs 43%) and improved survival at 1 year (49% vs 24%). However, a prospective study ³⁵ done by the Radiation Therapy Oncology group found median survival in the IORT group to be similar to that of conventional treatment.

Table III. Intraoperative radiotherapy as adjuvant treatment for pancreas cancer.

Series	Period	n	Regimen	Median survival (months)	Actuarial survival (%) 1 year	Actuarial survival (%) 2 year	Actuarial survival (%) 3 year	Actuarial survival (%) 5 years
Zerbi et al. 34	1985–1993	43	IORT (13–20 Gy)		71	24	7
		47	–		49 (p>0.05)	16 (p>0.05)	10 (p>0.05)
RTOG 35	1988–1990	51	IORT (20 Gy)	9		9 (18 months survival)
Reni et al. 36	1985–1988	127	IORT (10–25 Gy)	15.5				22
		76	–	12 (p=0.07)				6
Sindelar et al. 37		Total 32	IORT (20 Gy)	12
			–	12

Reni et al. ³⁶, in a retrospective study, allocated 127 patients to receive IORT following resection, while 76 had no adjuvant treatment after surgery. In addition to IORT, 56 patients received EBRT and 82 received chemotherapy (8 different regimens). Local recurrence occurred earlier in the group with no IORT (11 months) compared with the group that received IORT (median 14 months; p=0.02). There was no significant difference in survival. When data were further analysed, limited to patients with stage 1–2 cancer (30 in IORT arm and 19 in surgery arm), IORT significantly prolonged time to both local (12 months vs 17.5+ months; p=0.003) and distant failure (11.5 vs 17.5+ months; p=0.005) and overall survival (13 vs 18.5 months; p=0.01). The 5-year survival was 22±10% in the IORT group and 6±6% in the non-IORT group. Late complications of chronic upper abdominal pain and gastrointestinal bleeding were similar in both groups, although anastomotic stenosis/jejenal stenosis was only seen in the IORT group.

As most series on IORT are dogged by small numbers, inclusion of all stages of the disease and heterogenous treatment strategies, it is difficult to draw conclusions or make recommendations on IORT. The one small randomized trial on IORT ³⁷ was published in abstract form and found no difference in survival between surgery only and IORT (median survival 12 months in both groups). At the present time there is little evidence to support the use of adjuvant IORT, either alone or in combination with other forms of treatment.

Postoperative radiotherapy/chemoradiotherapy

In a prospective analysis of 14 patients receiving only postoperative radiation therapy, Bosset et al. ³⁸ demonstrated a median survival of 23 months, thus supporting the use of adjuvant treatment.

In a large multicentric prospective randomized trial (Table IV), Klinkenbijl et al. ³⁹ compared adjuvant chemoradiotherapy to no adjuvant treatment in both pancreatic and periampullary cancers. Between September 1987 and April 1995, 218 patients from 29 European centres were enrolled, 108 in the observation group and 110 in the treatment group. Patients in the radiotherapy group received split course radiotherapy with 40 Gy and concurrent 5-FU as continuous infusion, at a dose of 25 mg/kg per 24 hours, with a maximum daily dose of 1500 mg. In all, 207 patients were evaluable, 103 in the observation group and 104 in the treatment group. Median survival was 19 months in the observation group and 24.5 months for the treatment group (log rank p=0.208). In patients with pancreatic cancer, there was a larger difference in survival, being 12.6 months in the observation group and 17.1 months in the treatment group. Unfortunately the power of the study was rather weak and the difference in survival approached borderline significance (p=0.099). That this is likely to be a true result, rather than a type II statistical error is borne out by the results of the ESPAC-1 trial ¹⁰. Moreover, 68 (66%) patients in the observation arm and 67 (65%) in the treatment arm developed progressive disease. There was no advantage of adjuvant treatment in progression-free survival of either the entire group of eligible patients or in patients with pancreatic head cancer. This suggested that radiotherapy with 40 Gy does not prevent local recurrence and that radiotherapy and 5-FU after curative resection does not offer survival advantage, despite being safe and well tolerated.

Table IV. Adjuvant chemoradiotherapy: randomized controlled trials.

Series	Period	n	Regimen	Median survival (months)	Actuarial survival (%) 1 year	Actuarial survival (%) 2 year	Actuarial survival (%) 3 year	Actuarial survival (%) 5 years
Klinkenbijl et al. 39	1987–1985	110	40 Gy +5-FU	24.5	41
		108	–	19 (p=0.208)	51
ESPAC-1 interim 9	1994–2000	175	40 Gy +5-FU	15.5
		178	–	16.1 (p=0.235)
ESPAC-1 Final 10	1994–2000	145	40 Gy +5-FU	15.9		29		10
		144	–	14.8 (p=0.05)		41		20

5-FU, 5-fluorouracil.

In the ESPAC-1 trial ⁹, 70 patients were randomized to the chemoradiotherapy arm in the 2×2 factorial design, while a further 68 were randomly assigned to either chemoradiotherapy or no chemoradiotherapy. Chemoradiotherapy was administered as 20 Gy in 10 daily fractions over 2 weeks with 500 mg/m² 5-FU intravenously on days 1–3, repeated after 2 weeks. The median survival was 15.5 months in 175 patients who received chemoradiotherapy and 16.1 months in the 178 patients who did not receive chemotherapy (p=0.24). Thus there was no improvement in 2-year survival benefit achieved following chemoradiotherapy, and a confounding negative effect of chemoradiotherapy on the benefit of chemotherapy alone was apparent. In the final results of the ESPAC-1 trial ¹⁰, which only assessed patients enrolled in the 2×2 design, the median survival was 15.9 months among the 145 patients assigned to chemoradiotherapy and 17.9 months among the 144 patients who were not assigned to receive chemoradiotherapy (p=0.05). The estimated 5-year survival was 10% in the chemoradiotherapy arm compared with 20% in those who did not receive chemoradiotherapy (p=0.05). The median time to recurrence was 10.7 months in those assigned to chemoradiotherapy compared with 15.2 months in those who did not receive chemoradiotherapy (p=0.04). The lack of benefit of chemoradiotherapy is similar to that observed in the EORTC trial ³⁹ and one of the factors could be the delay in administering the follow-on chemotherapy in patients who received both treatments, thereby reducing the potential benefit of chemotherapy that is derived by administering it as soon as possible after resection.

Neoadjuvant treatment

Neoadjuvant (preoperative) chemoradiotherapy

Preoperative chemoradiation, by providing better tumour cell oxygenation, might have advantages over postoperative chemoradiation ⁴⁰. Moreover, with this modality, adjuvant therapy is commenced early, unlike in the postoperative setting where time to recovery from a major operation can delay the delivery of adjuvant treatment. It might also down-stage tumours, thus reducing positive surgical margins, and also avoid unnecessary laparotomy in patients with rapidly progressive disease ⁴¹. A criticism of preoperative chemoradiotherapy is that in those patients who progress on treatment, the opportunity for curative resection may have been missed.

In a prospective, multi-institutional phase II trial, Hoffman et al. ⁴² administered preoperative chemoradiotherapy to 53 patients with resectable pancreatic cancer: 5040 cGy radiotherapy was given, along with mitomycin 10 mg/m² on day 2 and 5-FU 1000 mg/m²/day continuous infusion on days 2–5 and 20–32. Forty-one patients underwent surgery, while 12 did not owing to death in 1 patient, toxicity in 1, local progression in 3, distant metastases in 6 and intercurrent illness in 1 patient. Of the 41 who were operated upon, 24 underwent resection while the rest had local extension/distant metastases that precluded resection. Median survival for the entire group was 9.7 months and for the patients who underwent resection, was 15.7 months. Treatment toxicity was primarily haematological, although a comparable number of patients suffered from biliary tract complications, i.e. obstruction or cholangitis.

Pendurthi et al. ⁴³ retrospectively analysed 25 patients with resected pancreatic carcinoma who received preoperative chemoradiotherapy versus 18 who received postoperative chemoradiotherapy. More, though not statistically significant, negative resection margins were observed in the preoperative chemoradiotherapy group (28% vs 56%), and also a significantly greater amount of fibrosis replacing the tumour in the preoperative chemoradiotherapy group (70% vs 40%; p=0.0001). There was no significant survival difference between the two groups (median 20 months vs 25 months; p=0.48), and no difference in toxicity and efficacy profile. However, 22% of patients intended for postoperative treatment did not receive treatment. Likewise, Spitz et al. ⁴⁴, in a comparative study between preoperative and postoperative chemoradiation, found that one-quarter of eligible patients in the postoperative chemotherapy group failed to receive treatment owing to prolonged recovery time. They found no difference in toxicity, patterns of tumour recurrence and survival between the two groups.

An Italian study ⁴⁵ assessed safety and efficacy of chemoradiotherapy given to both locally advanced (n=20) and resectable (n=8) pancreas cancer patients. Their protocol comprised 39.6 Gy EBRT and 5-FU given as continous infusion between days 1 and 4 at 1000 nmg/m²/day. Resected patients received 10 Gy IORT before reconstruction and, subsequently, six cycles of adjuvant chemotherapy with 5-FU, mitomycin C and adriamycin. Chemoradiotherapy was well tolerated in both groups, and the median survival in the resected group was 18.5 months.

In a prospective study, Magnin et al. ⁴⁶ studied 32 patients given preoperative chemoradiotherapy between November 1996 and December 2001. Radiation was given either as split course therapy of 30 Gy or standard fractionation therapy of 45 Gy over 5 weeks. Concurrent chemotherapy included continuous infusion of 5-FU and a bolus of cisplatin. All 32 patients completed the entire course of chemoradiotherapy, with 2 patients suffering grade 3 toxicities (weight loss and vomiting) and 1 patient died of grade 4 infection with febrile neutropenia. Nineteen patients underwent pancreatic resection, and one patient in this group developed a late toxicity of acute superior mesenteric artery thrombosis leading to death. Radiation-induced alterations in the blood vessels, although rare, are well known and have been attributed to several mechanisms: intimal fibrosis, the presence of fibrin and the accumulation of lipid-containing macrophages ⁴⁷. The 2-year overall survival rate in all 32 patients was 37.3%, with a median survival of 16 months. The 2-year overall survival rate in the 19 patients who underwent resection was 59.3%, with a median survival of 30 months. The 2-year disease-free survival rate was 43.7%.

In a later series from the same group, Moutardier et al. ⁴⁸ described their experience with preoperative chemoradiotherapy in patients with localized pancreatic head ductal adenocarcinomas. Among 87 patients, 17 underwent resection alone, 39 patients with potentially resectable cancers received preoperative chemoradiotherapy, as did 31 with locally advanced cancer. In the 39 potentially resectable patients, after completion of chemoradiotherapy, only 23 remained resectable at restaging. The median survival of patients who underwent chemoradiotherapy followed by resection was significantly higher than those who only underwent curative surgery (26.6 months vs 13.7 months; p≤0.005). There was a significantly high rate of biliary stent-related problems (23%) and two deaths due to superior mesenteric artery thrombosis. However, they did not experience high rates of chemoradiotherapy-related rehospitalization, unlike patients in some other series ⁴⁹,⁵⁰, nor did they experience anastomotic complications of pancreaticojejenostomies, and this could be related to the radiation-induced pancreatic fibrosis ⁴². On pathological analysis, 8 of the 23 resected patients had a major pathological response (35%) including two complete responses (9%). Long-term follow-up showed that none of the patients who received chemoradiotherapy followed by resection developed local recurrence.

In a study from the MD Anderson Cancer Centre, the overall median survival was 21 months, with a 10% local recurrence rate in 132 patients ⁵¹. In a phase II trial by Pipas et al. ⁵² in patients with resectable pancreas cancer, docetaxel and gemcitabine were given on days 1, 15 and 29, followed by EBRT with 50.4 Gy at day 43, together with gemcitabine twice weekly for 12 doses. In all, 24 patients with stage I–III disease were enrolled from January 2002. Nineteen patients completed treatment, with 11 responses (57%), including one complete radiological response. Fourteen patients underwent resection, with 11 of these having margin-negative resection, including 8 patients whose disease was felt to be unresectable or borderline unresectable prior to treatment. A dose finding study on neoadjuvant chemoradiotherapy with gemcitabine and accelerated hyperfractionated radiation in patients with potentially resectable pancreas cancer was conducted by Nakamori et al. ⁵³. Their recommended dose of gemcitabine was 800 mg/m² combined with 36 Gy accelerated hyperfractionated irradiation.

Neoadjuvant chemotherapy

In a phase II trial, Palmer et al. ⁵⁴ from Birmingham randomized 50 patients with resectable pancreas cancer to receive either neoadjuvant gemcitabine or gemcitabine with cisplatin. In this trial, 22/24 patients in the gemcitabine arm received a median of 85% of planned gemcitabine dose; 24/26 patients in the gemcitabine plus cisplatin arm received a median of 88% of planned gemcitabine and 92% of planned cisplatin dose. Grade III/IV haematological toxicity was experienced by nine patients in each group. Twenty-seven patients (54%) underwent resection; 18 patients had a bypass procedure and 5 patients had no surgery due to progressive disease. The 1-year survival from randomization was 46% in the gemcitabine arm and 61% for the gemcitabine and cisplatin arm. A phase III trial of neoadjuvant gemcitabine and cisplatin versus immediate surgery is planned.

Randomized controlled phase III trials assessing neoadjuvant therapy are lacking, therefore this cannot currently be endorsed as standard treatment.

Combination treatment

The rationale for the use of combination treatment (chemoradiotherapy followed by chemotherapy) as adjuvant treatment in resectable disease stems from a randomized controlled trial done by the Gastrointestinal Tumour Study Group (GITSG) in the USA ⁸. Forty-three patients were enrolled in this study, with 23 randomized to the adjuvant treatment arm and 21 to combined treatment in the form of split course EBRT (40 Gy) and concurrent 5-FU, followed by 5-FU for 2 years. The study was terminated prematurely, both because of a low rate of accrual and because of an increasingly large difference in survival between the study arms. The median survival for the adjuvant treatment group was 20 months, significantly longer than the 11 months in the no adjuvant treatment arm. However, five patients (24%) in the adjuvant treatment arm did not receive planned treatment owing to a prolonged postoperative recovery. Because there were so few cases, a further 30 patients were registered (not randomized) to the treatment arm and the median survival in this group was 18 months, with a 2-year survival rate of 46% ⁵⁵. Unfortunately, due to the small number of patients, the 95% confidence intervals of the survival curves were so large as to overlap with survival curves in patients receiving no additional treatment. Thus no convincing conclusion could be derived from this study. The lack of a concurrent control group in the follow-up study and inclusion of patients with body and tail cancers of the pancreas are other flaws of the GITSG studies.

Carducci et al. from John Hopkins initiated a phase II trial to assess the toxicities of abdominal/hepatic irradiation (liver 23.4 Gy; tumour bed 50.4 Gy) with concurrent chemotherapy (continous infusion of 5-FU and leucovorin), followed after a gap of a month by four cycles of chemotherapy (5-FU and leucovorin) in 14 patients who had undergone pancreaticoduodenectomy for periampullary or pancreatic head cancer ⁵⁶. Three patients developed grade 3/4 gastrointestinal toxicities during chemoradiotherapy, while toxicities in the subsequent chemotherapy-alone cycles were few. Abnormalities in liver function tests were noted in 50–60% of patients, but were asymptomatic. In this study, 12/14 (85%) patients progressed, with equal numbers of recurrences in the pancreatic bed and in the liver over the 12-month median follow-up. The median survival of patients was 417 days and similar to that of patients who refused the trial and received treatment as per the GITSG protocol.

Yeo et al. ³, in a prospective non-randomized single institution study, compared two different postoperative chemoradiation protocols to those of no adjuvant treatment. The chemoradiation protocols were either standard therapy (40–45 Gy EBRT with 5-FU, followed by weekly bolus 5-FU for 4 months) or intensive therapy (50.4–57.6 Gy EBRT to pancreas bed with 23.4–27 Gy prophylactic hepatic irradiation, given with and followed by infusional 5-FU plus leucovorin for 4 months). Between October 1991 and September 1995, 174 patients underwent pancreaticoduodenectomy, with 1 postoperative death. Ninety-nine patients opted for standard therapy, 21 for intensive therapy and 53 declined therapy. Patients in the postoperative adjuvant chemoradiation group had a significantly better survival of 19.5 months compared with the 13.5 months in the no adjuvant treatment group (p = 0.003). When comparison was made between the two chemoradiation protocols, there was no significant difference in survival for the intensive treatment group (median survival 17.5 months) compared to the standard therapy group (median survival 21 months; p=not significant). The limitations of this study are its non-randomized nature, their bias in favour of adjuvant treatment and the fact that many patients receiving standard therapy received it outside of the study institution and hence toxicity data could not be fully examined. Results from the Mayo Clinic also found an improvement in outcome with adjuvant treatment, but they experienced a high incidence of hepatic failure and peritoneal seeding ⁵⁷,⁵⁸.

The UK Pancreatic Cancer Trials Group (UKPACA) studied combined treatment in a total of 40 patients from 9 institutions between 1987 and 1993 ⁵⁹. All patients underwent pancreatoduodenectomy, of whom 34 had ductal adenocarcinoma and 6 had ampullary tumours. Radiation therapy consisted of a split course of 40 Gy separated by 2 weeks and 5-FU was administered as a radiosensitizer, just as in the GITSG regimen ⁸,⁵⁵. Two weeks after radiotherapy, 5-FU was given as a bolus once a week for a maximum of 24 weeks. No patient received the full course of chemotherapy and a median of eight treatments was given. The median survival for the 34 patients with pancreatic ductal adenocarcinoma was 13.2 months and the 5-year survival rate was 15%. Patients with positive lymph nodes had a significantly shorter survival time than those without lymph node involvement. A positive resection margin was associated with reduced survival only on multiple regression analysis once lymph node status was taken into account. In this study no patient died or required hospitalization because of 5-FU toxicity. It was felt that the optimum length of follow-on treatment with 5-FU should not be beyond 6 months in the adjuvant setting.

As gemcitabine is the recommended chemotherapeutic agent for advanced pancreatic cancer, it is increasingly being studied in the adjuvant treatment setting. Kachnic et al. ⁶⁰, in a prospective non-randomized study, assessed the feasibility of administering radiotherapy with concurrent 5-FU followed by 4 months of gemcitabine in both patients with resectable (n=9) and locally advanced (n=14) pancreas cancer. Adjuvant gemcitabine was well tolerated, with no incidence of radiation recall. A phase II study ⁶¹ evaluated the feasibility of postoperative administration of gemcitabine alone for three courses, followed by concurrent gemcitabine (at 300 mg/m²/week) and irradiation (40 Gy) in patients undergoing curative resection of pancreatic adenocarcinoma. All patients completed chemotherapy with gemcitabine, and all but two patients completed chemoradiotherapy. One patient developed grade 4 vomiting and thrombocytopenia while the other suffered disease progression. The median disease-free survival and overall survival was 6 and 15 months, respectively. The reasons for administering gemcitabine before combination therapy were that it could be started soon after surgery, tackle micrometastatic dissemination and select out patients with rapid disease progression. A phase I trial established that 39 Gy was the maximum tolerated dose when used with full dose gemcitabine after curative pancreatic resection ⁶².

A prospective non-randomized trial from Germany studied concurrent chemoradiotherapy with gemcitabine and cisplatin, followed by two cycles of cisplatin and gemcitabine, in patients with R1 resection status ⁶³. Haematological toxicities were the most common side effects in the 30 patients enrolled. The median progression-free survival was 10.6 months, and the median overall survival was 22.8 months. Distant metastases occurred in 14 patients, and only one patient had local failure during a median follow-up of 15 months. The true benefit of this treatment strategy is being assessed in an ongoing randomized trial.

The Radiation Therapy Oncology Group Study no. 97-04 is a phase III study of pre-and post-chemoradiation 5-FU (for 3 months at a dose of 250 mg/m²/day) versus pre- and post-chemoradiation gemcitabine (for 3 months at a dose of 1000 mg/m²/day) for postoperative adjuvant treatment of resected pancreatic adenocarcinoma. Accrual of >500 patients has been completed and the results are awaited.

Meta-analysis

Stocken et al. ¹⁴ performed a meta-analysis investigating the roles of adjuvant chemoradiation and chemotherapy following resection of pancreas ductal adenocarcinoma on survival. Individual patient data were available in four (875 patients) of the five selected randomized controlled trials (total number of patients with pancreatic adenocarcinoma = 939). Assessment of adjuvant chemotherapy trials revealed a 25% significant reduction in the risk of death (hazard ratio = 0.75, CI: 0.64, 0.90, p_strat=0.001). The median survival was 19 months with chemotherapy and 13.5 months without. On the other hand, there was no significant difference in the risk of death with chemoradiation (hazard ratio = 1.09, 95% CI: 0.89, 1.32, p_strat=0.43). The median survival was 15.8 months with chemoradiation and 15.2 months without. On subgroup analysis, chemoradiation was more effective (χ²=4.2, p=0.04) and chemotherapy less effective (χ²=7.3, p=0.007) in patients with a positive resection margin. These results provide strong evidence for institution of adjuvant systemic chemotherapy following curative surgery, although further RCTs may help decide the optimal chemotherapeutic agent. Although not recommended as a standard form of adjuvant treatment, chemoradiation needs to be studied further in the patients with positive resection margins.

Conclusions

There is a definite role for adjuvant treatment of pancreatic cancer following resection, from the available literature. There is controversy on either side of the Atlantic over the form of treatment, i.e. chemotherapy alone or combination therapy. A meta-analysis ¹⁴ of randomized adjuvant therapy trials concluded that chemotherapy is effective as adjuvant treatment, but not chemoradiotherapy. It highlighted the need for further studies with chemoradiation in patients with R1 resection. The outcome of the ESPAC-1 trial firmly supports the use of adjuvant chemotherapy. The ESPAC-3 trial should answer the question of the agent to be used in adjuvant chemotherapy, i.e. gemcitabine or 5-FU. Currently there does not seem to be a place for IORT alone in the adjuvant setting. The rationale for combination therapy is based on the results of the small GITSG study, with its many shortcomings. The results of the RTOG 97-04 will answer the question on the agents to be used and will assess results with more modern radiation techniques, but will still not help in resolving the controversy on chemotherapy versus combination therapy. There are currently no results from large RCTs to support the routine use of either regional therapy or neoadjuvant treatment. However, the encouraging results following regional therapy have prompted the ESPAC-2 trial, which will shed further light on this subject. The use of novel agents in combination with current modalities is being tested in phase II/III trials. Large co-operative group trials can help develop better and more effective therapies to improve the outlook and quality of life of patients with this devastating disease.