Current state of adjuvant therapy in resected pancreatic adenocarcinoma

Abstract

Pancreatic carcinoma cannot generally be cured by surgery alone. This review summarizes the development of adjuvant therapy over the past two decades. Four randomized controlled trials compare long-term survival of different treatments. The small GITSG-study supports combined chemoradiation, but the EORTC-study found no significant effect. A Norwegian study of adjuvant chemotherapy found an increased median survival, but no effect beyond two years. The large ESPAC-1 study shows a benefit for 5-FU based chemotherapy, while chemoradiation had a negative effect. Thus, evidence favours adjuvant therapy, but 5-FU may not be the ultimate drug. Support for gemcitabine is given by preliminary data from a German randomized trial, and further American and European studies are upcoming. However, postoperative therapy is problematic, as 20–30% of resected patients never undergo treatment because of slow recovery or other reasons. Preoperative therapy has some theoretical advantages, and moreover, patients with rapidly progressive disease may be spared surgery. Randomized controlled trials are lacking, but published results compare well with postoperative, adjuvant therapy. The value of locally targeted therapy is difficult to assess. Reasonable results have been obtained with regional chemotherapy, whereas intraoperative radiotherapy does not seem to increase survival despite reducing reducing local recurrences.

Despite many years of continous efforts, pancreatic carcinoma remains a highly lethal disease and a challenge to the medical community. Surgery has developed considerably through the years, and the postoperative mortality reported by the leading cancer hospitals now amounts to a few percent [1–7]. Correspondingly, 5-year survival in radically resected patients has improved, and figures in the range 15–25% have been reported [2], [6–12]. However, death continues beyond five years, and in reality the absolute majority of resected patients will die from the disease [7], [8], [12]. Extensive resections with vessel resection and clearance of retroperitoneal lymph nodes and soft-tissue can be done without significantly increased postoperative mortality, but have not been shown to improve survival [5], [13–15]. Thus, refinement of pancreatic surgery has been unable to change the fact, that locoregional recurrence occurs in 50–80% of resected patients, and a similar number will develop liver metastases [16]. During the past decades a variety of adjuvant chemotherapy and/or radiotherapy have been tried in order to achieve better survival for the operated patients.

This paper aims at a review of adjuvant therapy published in English during the years 1984–2005.

Methods

A Medline search of studies published in English during the years 1984–2005 was made using the search terms: “pancreas/pancreatic, cancer/carcinoma, chemotherapy, radiotherapy, chemoradiotherapy, adjuvant, neoadjuvant, preoperative, intraoperative” in various combinations. A further search of the reference lists of all retrieved articles and reviews were done to ensure that all possible relevant studies were included. In the review were included original studies concerning protocol-based treatment of at least 20 patients, and including long-term results (at least two years). The reports were classified according to the dominant principle of treatment, although it is realized that many protocols are heterogenous and include treatment combinations. Thus, the studies were grouped as:

1. chemoradiation (Table I);

2. systemic chemotherapy, incl. ESPAC-1 (Table I);

3. regional chemotherapy (Table II);

4. neoadjuvant/preoperative chemoradiotherapy (Table III);

5. intraoperative therapy (Table IV).

Results of literature review

1) Postoperative chemoradiation

Adjuvant therapy was developed on the basis of studies, which indicated – contrary to common wisdom at the time – that 5-fluorouracil (5-FU) as well as radiation had a demonstrable effect in advanced pancreatic cancer [17–21]. In 1985, Gastrointestinal Tumor Study Group (GITSG) published a randomized controlled trial of postoperative adjuvant therapy in resected pancreatic cancer, using a regimen of two courses of 20 gray (Gy) combined with bolus injections of 5-FU (500 mg/m²/3days), followed by weekly 5-FU for two years or until recurrence occurred [22]. The observed median survival was 20 and 11 months in the treatment (n = 21) and control arms (n = 22), respectively, resulting in a significantly better two-year survival with active treatment (43% vs. 18%). Criticism has focussed on small sample size, slow accrual of patients, and the premature ending of the study. Bias was also introduced in that patient recovery and performance status decided if patients could receive planned radiotherapy. Twenty-four percent (5/21) of the patients could not begin planned chemoradiation until more than ten weeks after pancreatic resection because of prolonged recovery. This rate of drop-out has been a common phenomenon also in later studies. A confirmatory, non-randomized study of 30 patients receiving the same adjuvant regimen showed similar results: median survival 18 months, two-year survival 46% [23]. The control group from the first report was used, but the patients in the confirmatory study had better performance status, which is known to influence survival [24].

In the coming years, these studies have had great impact on pancreatic cancer management in USA, while Europeans have been more hesitant. After the GITSG studies, several non-randomized studies of 5-FU-based chemoradiation have been published (Table I). In a non-randomized study from Philadelphia [25], historical controls receiving surgery alone (n = 33) were compared with two regimens of chemoradiation. A median survival of 15–16 months were observed in all three groups. Patients in the mitomycin C -group had 2- and 3-year survivals of 43% and 34%, respectively, compared with 30% and 5% with radiotherapy and 5-FU, and 35% and 8% with surgery alone. However, treatment groups were small and probably not comparable regarding prognostic parameters.

Table I.  Adjuvant chemoradiation and adjuvant chemotherapy.

			Surgical radicality (%)			Actuarial survival (%)
Study	Treatment	No. of patients	RO	NO	Median survival (months)	2-y.	3-y.	5-y.	Comments
Kalser (22)	observation	22	100§	–	10.9	18	–	–	RCT
	Rx + F	21			21.0	43		–
GITSG (23)	Rx + F	30	100	–	18	46			supportive study
Whittington (25)	observation	33	72§	75–78§	15	35	8	–
	Rx + F	19			15	30	5	–
	Rx + F M	20			16	43	34	–
Foo (26)	control	101	–	–	16.5	–	9	4
	control	89			12	25	11	7
	Rx + F	29			22.8	48	24	12
Bakkevold (32)	observation	31	45–70§	–	11	32	30	8	RCT, incl. ampullary neoplasms
	F + A+M	30			23	43	27	4
Yeo (2)	control	53	79	23	13.5	30	–	–
	Rx + F (st.)	99	65	27	21.0	44	–	–
	Rx + F+L (int.)	21	81	24	17.5	22	–	–
Abrams (27)	Rx + F	23	–	–	15.9	30	20	–
Paulino (28)	Rx + F	38	74	53	21	–	16	–	Different protocols used.
Klinkenbijl (30)	observation	104	76§	61§	19.0	41	–	22	RCT, incl. ampullary neoplasms
	Rx + F	103			24.5	51	–	28
Sohn (4)	observation	132	70§	73§	11	30	–	9
	Rx + F (±M±D)	366			19	39	–	20
Mehta (29)	Rx + F	52	65	41	32	62	39	–
ESPAC-1 (36)	observation	69	81–84§	50–58§	16.9	–	–	11	RCT
	Rx + FL	73			14.2	–	–	13
	Rx	73			13.9	–	–	7
	FL	75			21.6	–	–	29

Treatment: Rx: radiotherapy; A: adriamycin; C: cisplatinum; D: dipyridamol; F: 5-flurouracil; IFN: interferon-alpha; L: leukovorin; M: mitomycin C; Mo: mitoxantrone; S: streptozotocin; st.: standard protocol; int.: intensive protocol.

Surgical radicality: Reported proportion of study patients with negative resection margins (RO) and negative lymph nodes (NO). §: refers to all subgroups together.

RCT: randomized controlled study.

From the Mayo clinic, Foo et al. [26] reported a series of 29 patients treated with a median dose of 54 Gy (range 35–60 Gy) radiotherapy, which was mostly combined with two courses of 5-FU (500 mg/m²/3days). The median survival of 23 months and the two-year survival of 48% were significantly better than in a historical series treated with surgery alone.

In 1995, Yeo reported experience at the Johns Hopkins University with 5-FU based adjuvant chemoradiation in 56 patients [24], and in the subsequent expanded study [2] three options of postoperative treatment were compared: 1) no postoperative adjuvant therapy; 2) standard therapy consisting of 40–45 Gy radiotherapy combined with two 3-day courses of 5U; 3) intensive therapy consisting of 50–58 Gy to the pancreatic bed and 23–27 Gy to the liver combined with weekly 5-FU and leukovorin for four months. Standard treatment resulted in a median survival of 21 months and 2-year survival of 44%, compared with 17.5 months and 22% with intensive treatment. The control group without postoperative adjuvant treatment had a median survival of 13.5 months and a 2-year survival of 30%. However, there were significant differences in age, postoperative complications, reoperations and length of stay between the treatment groups.

Also from Johns Hopkins University, the effects of radiotherapy combined with 5-FU and leukovorin were studied [27]. Toxicity was mild and manageable. The over-all median survival was 14 months. Patients with pancreatic cancer (n = 23) had a median survival of 16 months, and 2- and 3-year survivals 30% and 20%, respectively. However, the authors conclude that the used regimen had low effect, and that improved or additional therapy will be required.

In 1999, Paulino et al. reported results of adjuvant chemoradiation in 38 patients [28]. Patients received 50 Gy to the pancreatic region, and in 30 patients, 5-FU was given during radiotherapy. The median survival was 21 months, and 1- and 3-year survivals were 71% and 16%, respectively. The median survival in 30 patients treated with both radiotherapy and 5-FU was 26 months. However, this study concerns patients treated in 1978–97, and is not obvious from the paper what changes in treatment that may have occurred during these 20 years.

In 2000, Mehta reported results in 52 patients receiving chemoradiation with concurrent protracted venous infusion of 5-FU (200–250 mg/m²/d) [29]. All patients were prescribed 25×1.8 Gy, and those with positive resection margins were given extra radiotherapy up until 54–60 Gy. In addition, eight patients were given 10–15 Gy of intraoperative radiotherapy (IORT). The treatment was well tolerated and only grade II–III toxicities were noted, mainly as neutropenia, anemia and weight loss. Ninety five percent of patients were able to complete the treatment course. The patients had a median survival of 32 months, and actuarial survival at two and three years was 62% and 39%, respectively.

A recent, comprehensive study from Johns Hopkins University patients reported results in 498 patients, which were treated either with pancreatic resection alone (n = 132), or resection followed by various protocols of chemoradiation (n = 366) [4]. Patients receiving adjuvant therapy had a median survival of 19.5 months and a two-year survival of 39%, which was significantly better than in untreated patients (13.5 months, 30%, p = 0.003). The more intensive regimen did not appear to improve survival. From a surgical point of view, this large material had an impressing postoperative mortality of 2.3%, and a prognostic importance of resection margin status, nodal status, and tumour size was shown. Regarding adjuvant therapy, however, an inherent bias was that the decision to deliver therapy was influenced by patient choice, age, slow recovery, and poor performance status.

In 1999, EORTC (European Organisation for Research and Treatment of Cancer) published a multicentre trial of adjuvant chemoradiation in pancreatic and periampullary cancer [30]. One group of patients was randomized to two courses of radiotherapy in combination with infused 5-FU, whereas the control group received surgery alone. After exclusion of ineligible cases (n = 11), there remained 104 patients in the treatment group and 103 patients in the control group. The median survival was 24.5 months with treatment group compared with 19 months for the control group (p = 0.208). The two-year survival was 51% and 41% in the treatment and control groups, respectively. In the subgroup of pancreatic cancer the median survival was 17 months with treatment vs. 13 months for controls, and the corresponding two-year survival estimates were 34% vs. 26% (p = 0.099). A bias was that as many as 22% of patients in the treatment arm did not receive planned treatment. The study has also been criticized regarding deficient stratification for resection margin status, quality assurance and abscence of follow-on chemotherapy. Furthermore, the study included pancreatic as well as periampullary cancer, and it was insufficiently powered for subgroup analysis. Some critics have described it as an “underpowered positive trial” [31].

2) Postoperative, systemic chemotherapy

In a Norwegian multicentre trial of cancer of the pancreas and papilla of Vater [32], 61 radically resected patients were randomized to 5-FU ( 500 mg/m²), doxorubicin (40 mg/m²), mitomycin C (6 mg/m²), or into a control group (n = 31). The treatment group had significantly longer median survival than the controls (23 months vs. 11 months; p = 0.03), but survival after two years did not differ. Criticism has been raised against slow accrual of patients, and the inclusion of cancers of the papilla, which generally have a better prognosis.

The European Study Group of Pancreatic Cancer (ESPAC) has conducted the largest randomized trial of adjuvant chemoradiation and chemotherapy (ESPAC-1). In a joint effort between 53 hospitals in 11 countries, 289 patients were randomised into a 2×2 factorial study, which set out to compare postoperative chemoradiation, postoperative chemotherapy, both treatments or none. Preliminary reports were published after a median follow-up of 10 months [33–35], and in 2004, the long-term results of the study were published after a median follow-up of 47 months [36]. The study found a significant benefit of adjuvant chemotherapy; the median survival being 20.1 months among 147 patients who received chemotherapy (± chemoradiation) vs. 15.5 months among 142 patients without chemotherapy (± chemoradiation). In contrast, 145 patients receiving chemoradiation (± chemotherapy) had a survival disadvantage-median survival 15.9 months- compared with 17.9 months among 144 patients without chemoradiation (± chemotherapy). Subgroup analyses (for which the study was underpowered) showed median survivals of 16.9 months in observation group, 12.2 months with chemoradiation, 21.6 months with chemotherapy, and 19.9 months with both chemotherapy and chemoradiation. The corresponding five-year survival estimates were 11%, 7%, 29% and 13%, respectively. The study group thus recommended “that standard care for patients should consist of curative surgery followed by adjuvant systemic chemotherapy” [36].

In an accompanying, critical editorial it was pointed out, that although the two-by-two factorial design makes it possible to explore two independent treatments in the same study, a confounding factor is introduced by the sequential nature of the two therapies, i.e. treatment-related toxicity of the first therapy will interfere with the possibilty to deliver the second therapy [37]. In a recent editorial [38], further criticism of the lack of details regarding radiation fields, machine and dosimetry specifications and overall quality assurance of the radiotherapy has been raised. They consider the “split course” radiotherapy obsolete, and moreover, they are concerned about the fact that 30% of the chemoradiation patients received a non-uniform dose or no radiotherapy at all. These deficiencies may have contributed to treatment-related toxicity, and may have been the cause of the decreased survival in the chemoradiation group [38]. Another bias was introduced by the fact that patients were able to start chemotherapy 48 days (mean) after surgery, compared with 61 days for chemoradiation. This raises the question as whether the patients in the chemoradiation arm had a worse performance status. Non-adherence was common also in the chemotherapy group; one third of the patients did not complete the regimen and 17% received no chemotherapy at all. It is pointed out that the non-adherence is unexplained, and the authors warn against premature conclusions and therapeutic nihilism [38].

Besides these studies, the early results of a randomized trial presented at the 2005 ASCO meeting need to be mentioned. The study concerns 368 patients, who after pancreatic resection received either Gemcitabine (1000 mg/m2, 3 times monthly for 6 months) or no active treatment [39]. In the preliminary analysis, gemcitabine-treated patients had a better disease-free survival than controls (DFS 14.2 vs. 7.5 months). The treatment was well tolerated with reasonable toxicity. The final results are expected in late 2005.

3) Regional chemotherapy

The rationale behind regional chemotherapy is to deliver high concentrations of the active agent to the major targets of recurrence, i.e. mainly the liver bed, while limiting dose-related, systemic toxicity. The chemotherapy is usually given through catheters placed in coeliac/hepatic artery or portal vein. Early reports of intraarterial adjuvant therapy were published by Ishikawa [40], [41]. In these studies, catheters were placed in the main tributaries of the coeliac trunk and portal vein and used for 28–35 days postoperatively to administer 5-FU (125 mg/m² daily). In the long-term follow-up, 1-, 3- and 5-year survivals were 92%, 54 and 39% [40], [41]. It is stated that the survival was significantly better than in two groups of mixed historical controls. However, it is not clear if the groups are comparable regarding prognostic indicators as nodal and resection margin status. Noteworthy in this study is the significantly lower rate of death from hepatic metastases, 8%, with intraarterial adjuvant therapy compared with controls, and the authors suggest that this effect improved survival.

Further studies have been published by Beger et al. [42], [43] using intraarterial chemotherapy through a transcutaneous catheter via the femoral artery. In the long-term follow-up [43], this treatment resulted in a mean survival of 23 months and 3-year survival of 48%. In the control group, the mean survival was a mere 10.5 months. It is stated that this represents a significant difference (p < 0.001), but the control group seems to have included less patients with RO-resection. In a separate analysis of RO-resected patients, a significantly better survival was found with RO-resection together with intraarterial treatment compared with historical controls receiving RO-resection alone. The calculated 1-, 2-, 3-year survivals were 95%, 59% and 43% in 19 patients with active treatment compared with 50%, 10% and 5% with RO-resection alone.

Ozaki [44] has reported a study of multimodality treament of resected pancreatic adenocarcinoma, which used a combination of 20–30 Gy IORT together with intraoperative hepatic artery or portal vein infusion of 6–10 mg mitomycin C. Eight patients survived more than five years, three of them more than ten years. Many of the patients were in an advanced stage of disease, and the authors state that their treatment “appeared to have a benefit for the prognosis”.

In summary, studies of regional chemotherapy show median survivals of 23–42 months and 3-year survivals of 39–48%, compared with 10–24 months and 12–29%, respectively, with surgery alone. The therapy seems safe, but the intravascular catheters have caused some problems, including intimal damage in 2% and one case of pseudoaneurysm [42].

More evidence of the benefits of this therapy is obviously needed, as only a limited number of relatively small, non-randomized studies exist (Table II).

Table II.  Regional chemotherapy.

			Surgical radicality (%)			Actuarial survival (%)			Comments
Study	Treatment	No. of patients	RO	NO	Median survival (months)	2-y.	3-y.	5-y.
Ishikawa (41)	control/F	67	81	25	24	–	39	27
	(reg)F	27	81	19	42	–	54	39
Beger (43)	control	25	100§	–	10.5	–	12	–
	C + F+L + Mo	26			23	–	48	–
Ozaki (44)	M + IORT	27	81§	–	30	–	46	31	IORT (20-)30 Gy

Treatment: (reg): regional chemotherapy; IORT: intraoperative radiotherapy; A: adriamycin; C: cisplatinum; D: dipyridamol; F:5-flurouracil; IFN: interferon-alpha; L: leukovorin; M: mitomycin C; Mo: mitoxantrone; S: streptozotocin.

Surgical radicality: Reported proportion of study patients with negative resection margins (RO) and negative lymph nodes (NO).

§ refers to all subgroups together.

4) Neodjuvant/preoperative chemoradiotherapy

Early studies demonstrated an effect of preoperative radiotherapy alone in pancreatic cancer [18], [19], [21], [45]. In the further development, protocols combining preoperative radiotherapy and 5-FU based chemotherapy were designed [46–49] (Table III).

Table III.  Neoadjuvant therapy.

			Surgical radicality (%)			Actuarial survival (%)
Study	Treatment	No. of patients	RO	NO	Median survival (months)	2-y.	3-y.	5-y.	Comments
Staley (16)	(Pre)Rx + F	39	82	62	19	33	25	19	IORT 10 Gy
Spitz (50)	(Pre)Rx + F	41	88	54	19.2	39	35	28	IORT 10–15 Gy
	(Post)Rx + F	19	74	37	22.0	52	40	40
Pendurthi (51)	(Pre)Rx + FM	25	72	72	20	40	20	–
	(Post)Rx	18	44	6	25	40	20	–
Hoffman (52)	(Pre)Rx + FM	24	71	81	15.7	27	–	–
Pisters (53)	(Pre)Rx + F	35	90	35	25	60	23	–	IORT 10–15 Gy
Snady (54)	controls	28	–	–	10.5	17	11	–	Different protocols used.
	(Pre)Rx + FSC	20	–	–	32.3	61	32	–
	(Post)Rx + F	63	–	–	16.2	34	13	–
Breslin (55)	(Pre) Rx + F/P/G	132	88	52	21	–	–	23	IORT 10–15 Gy
White (59)	(Pre)Rx + F/M/C	39	72	70	>16	32	–	28	Different protocols used.
Pisters (56)	(Pre)Rx + P	20	68	47	19	–	28	–	IORT 10–15 Gy

Treatment: Rx: radiotherapy; A: adriamycin; cisplatinum; D: dipyridamol; F: 5-flurouracil; IFN:interferon-alpha; L: leukovorin; M: mitomycin C; Mo: mitoxantrone; S: streptozotocin; (Pre): preoperative treatment; (Post): postoperative treatment; IORT: intraoperative radiotherapy.

Surgical radicality: Reported proportion of study patients with negative resection margins (RO) and negative lymph nodes (NO).

In 1996, Staley et al. at the MD Anderson Cancer Center reported results for a group of 39 patients, who completed 30–50 Gy of preoperative radiotherapy concurrent with infused 5-FU before pancreatic resection [16]. A subgroup of 33 patients was also given 10 Gy IORT to the pancreatic bed. The median survival was 19 months, and the two-year survival was 33%.

In a later study from the same institution, Spitz [50] compared results with the same neoadjuvant treatment in 41 patients with a group of 19 patients treated with postoperative chemoradiotherapy. Both the pre- and postoperative groups received 10–15 Gy IORT to the pancreatic bed. Actuarial survivals at 2, 3 and 5 years were 39%, 35% and 28% in the neoadjuvant group compared with 52%, 40% and 40% in the group receiving postoperative therapy.

In 1998, Pendurthi et al. from Philadelphia compared results with preoperative chemoradiotherapy with the results of postoperative, adjuvant treatment [51]. Preoperative treatment resulted in treatment delaying greater than one week in 2/25 (8%) patients. On the other hand, 22% (5/23) of patients scheduled for postoperative chemoradiotherapy did not receive treatment, either because of delayed postoperative recovery or patient noncompliance. These patients were excluded from the survival analysis. There were no significant survival differences between the preoperative and postoperative treatment groups (median survival 20 months vs. 25 months).

Also from Philadelphia, Hoffman et al. reported results with the same regimen of preoperative chemoradiotherapy in 53 patients with localised adenocarcinoma of the pancreas [52]. After restaging and exclusion of progressive disease or otherwise unfit patients, pancreatic resection was carried out in 24 patients. In resected patients, a median survival of 16 months and a two-year survival of 27% were noted.

In an effort to decrease treatment-related toxicity, Pisters et al. developed preoperative rapid-fractionation treatment delivered on an outpatient basis over two weeks as 10×3 Gy concurrent with infused 5-FU [53]. The regimen was well tolerated; hospital admission was necessary in one patient only. Thirty-five patients started treatment, but because of medical complications or progressive disease only twenty of them were finally resected. Furthermore, the pancreaticoduodenectomy was in 50% (10/20 cases) combined with resection of major vessels. An extra 10–15 Gy of IORT was given to the pancreatic bed. With this strategy, a median survival of 25 months and two- and three-year survivals of 60% and 23% were achieved.

An extensive protocol of preoperative chemoradiotherapy was reported by Snady et al. [54], who treated 20 patients with 54 Gy of radiotherapy combined with a regimen of 5-FU, streptozotocin and cisplatin. Preoperative chemoradiotherapy combined with pancreatic resection resulted in a median survival of 32 months, and the 2-year and 3-year survival estimates were 61% and 32%, respectively. With postoperative treatment, the median survival was 16 months and and 2-year and 3-year survival was 34% and 13%, respectively. With surgery alone, the corresponding figures were 10.5 months, 17% and 11%. However, the groups are obviously selected differently and do not seem comparable.

In 2000, a comprehensive study of 132 consecutive patients who received neoadjuvant chemoradiotherapy followed by pancreaticoduodenectomy for adenocarcinoma of the pancreatic head at the MD Anderson Cancer Center was published by Breslin [55]. The overall median survival was 21 months, and the 5-year survival was 23%. In the multivariate analysis, only female sex and negative lymph node status were associated with improved survival duration.

In a later publication from MD Anderson Cancer Center, Pisters reported a larger series of 37 patients receiving neoadjuvant therapy, 10×3 Gy over two weeks, together with weekly Paclitaxel 60 mg/m² for three weeks [56]. After completion of the neoadjuvant chemoradiotherapy, pancreatic resection was ultimately done in 20 patients. At surgery, thirteen (65%) patients were given 10–15 Gy of IORT (IORT was omitted in six patients subjected to combined pancreatic resection and vascular resection). The median survival in this group of patients was 19 months, and three-year survival was 28%. Compared with 5-FU-based protocols, patients receiving paclitaxel experienced more treatment-related toxicity. Grade 3 toxicity was noted in 46% (16/35) of patients; seven of them had grade 3 nausea and vomiting, requring hospital admission in four patients.

A group at the Duke University has also studied preoperative chemoradiotherapy, and its “down-staging” effect [57–59]. In an extensive study from 2001, 111 patients with pancreatic cancer were staged by pre-treatment CT-scanning as having either potentially resectable (n = 53) or locally advanced disease (n = 58) [59]. After subsequent delivery of chemoradiotherapy, CT-restaging was performed. Surgical exploration was performed in patients judged fit for surgery, who had stable or potentially resectable disease. Pancreatic resection was possible in 11/58 patients with “locally advanced tumour” and 28/53 patients with “potentially resectable tumour” on the first CT-scanning. The median survival in the resected patients exceeded 16 months with actuarial 1-, 2- and 5-year survival rates of 80%, 32% and 28%. Resection in the patients intially judged as “locally advanced” (n = 11) could mostly be limited to “standard” pancreatic resection; portal vein resection was required in one patient only. With this strategy, the surgical margins were negative in 73% and lymph nodes in 60%, which was comparable to the findings reported for the “potentially resectable tumours” (RO 71%, NO 70%).

5) Intraoperative radiotherapy

IORT has the potential to deliver high doses of radiation to sites prone to develop locorogional recurrence, while radiosensitive normal tissue, i.e.intestines, may be displaced and shielded. IORT was first used for palliative purposes in advanced, unresectable pancreatic cancer, but does not seem to confer any survival benefit in this situation [60]. Several later reports of IORT have focussed on the adjuvant use after pancreatico-duodenectomy (Table IV shows studies with n ≥ 20). IORT has also been used together with chemoradiotherapy [16], [50], [53], [55], [56] (Table III)

Table IV.  Intraoperative radiotherapy.

			Surgical radicality (%)			Actuarial survival (%)
Study	Treatment	No. of patients	RO	NO	Median survival (months)	2-y.	3-y.	5-y.	Comments
Zerbi (61)	control	47	66	–	12	24	7	–
	IORT	43	60.5	–	19	16	10	–
Fossati (62)	Control	44	66	–	12.5	16	10	–
	IORT	32	69	40	19	26	9.9	–
Di Carlo (63)	control	47	–	–	14				Chemotherapy also given
	Adj.+IORT	54	–	–	17
Hosotani (64)	control	64	58	31	7.9	8	8	4
	(Post)Rx/IORT	86	67	46	12.8	14	8	8
Sunamura (65)	control	24	–	–	16.7	–	21	–
	IORT	20	–	–	15	–	18	–
Kokubo (66)	control	39	71	41	11	–	16	–	Chemotherapy also given
	IORT	34	95	–	15	–	25	–
	IORT + Rx	18	–	–	17	–	24	–
Alfieri (67)	control	20	90	55	10.8	–	–	6
	IORT	26	90	50	14.3	–	–	16
Reni (68)	control	19	80§	–	13	28	6	6	Stage I–II subgroup
	IORT	30			18.5	50	28	22

Treatment: Rx: radiotherapy; A: adriamycin; cisplatinum; D: dipyridamol; F: 5-flurouracil; IFN: interferon-alpha; L: leukovorin; M: mitomycin C; Mo: mitoxantrone; S: streptozotocin; (Pre): preoperative treatment; (Post): postoperative treatment; IORT: intraoperative radiotherapy.

Surgical radicality: Reported proportion of study patients with negative resection margins (RO) and negative lymph nodes (NO).

§ refers to all subgroups together.

In a report from Milan, Zerbi [61] compared 43 patients treated with pancreatic resection and IORT with 47 patients receiving surgery alone. The number of pre- and postoperative complications did not differ, so IORT seems safe. Patients given adjuvant IORT had a median survival of 19 months, and two- and three-year survival of 16% and 10%. vs. 12 months, 24% and 7%, respectively with surgery alone (statistically non-significant). In a detailed analysis, IORT seemed to reduce the rate of local recurrences from 56% to 27%.

In a subsequent study from Milan, Fossati [62] reported on 33 patients treated with surgical resection and IORT. After exclusion of one operative death, 32 patients were treated either with IORT alone (n = 20) or a combination of IORT, EBRT and chemotherapy (n = 12). Group I treated with surgery and IORT alone had a median survival of 18 months, vs. 20 months when chemotherapy was added. However, deviations from the complete chemotherapy protocol seem to have been common.

Di Carlo [63] compared 54 patients given adjuvant IORT with 47 patients treated with pancreatic resection without adjuvant therapy. The median survival was 17 months in the IORT group compared with 14 months with surgery alone. On subgroup analysis, IORT in combination with chemotherapy and radiotherapy was associated with signficantly better survival. Also in this study, local recurrences were less frequent with (38%) than without IORT (54% local recurrences).

From Kyoto, Hosotani [64] reported 86 patients treated with resective surgery and adjuvant radiotherapy (both IORT and EBRT), and compared with a group receiving surgery alone (n = 64). Altogether 45 patients received IORT, either in combination with postoperative radiotherapy (n = 31) or alone (n = 14). Another group of 37 patients received postoperative radiotherapy alone. The authors suggested that IORT may give a survival benefit of approximately five months (13 vs. 8 months). Heterogenous treatment and inclusion of patients in all stages (I–IV) of disease make this non-randomized study difficult to assess.

A negative effect of IORT was reported by Sunamura [65], who found that 20 patients treated with IORT had poorer survival (15 months, three-year survival 18.2%) than 24 patients treated without IORT (17 months; 21%). None of the findings were statistically significant, and, data on tumour stage, resection margins and nodal status are lacking.

In 2000, Kokubo [66] reported results with IORT in 34 patients, either alone (n = 16) or in combination with postoperative radiotherapy (n = 18). With combination therapy median survival and two-year survival were 17 months and 24% compared with 15 months and 25% for surgery and IORT and 11months and 16% for the controls receving only surgery. Also this study material is heterogenous (stage III–IV). Further, 21 patients received additional, regional chemotherapy, which was found to increase survival.

In a report from Rome [67], 26 patients treated with pancreatic resection, IORT and EBRT were compared with 20 patients without adjuvant therapy. With active treatment, the rate of local recurrence was decreased (51% vs. 71%, p > 0.01). There were statistically non-significant increases in median survival and 5-year survival.

In a recent study from Milan [68], 127 patients were treated with IORT. In a subgroup analysis, 30 patients in stage I–II treated with surgery and IORT were compared with 19 patients treated with surgery alone. Medial survival was 18.5 months and two- three- and five-year survivals were 50%, 28% and 22% compared with 13 months and 28%, 6%, 6% in the control group. Altogether, it was concluded that IORT did not increase operative morbidity or mortality, while giving a significant survival benefit. A significant reduction of local recurrence (15% vs. 33%) was also noted.

In summary, IORT may reduce local recurrence by half [61], but long-term survival does not seem to be increased (Table IV). The most plausible reason is that IORT, like other locoregional therapies, does not adequately take care of the distant metastases, especially to the liver. Also, the value of adding IORT to neoadjuvant therapy is difficult to assess (Table III) [16], [50], [53], [55], [56].

Discussion

Adjuvant therapy has evolved through a large number of studies, but to date only four randomized controlled trials compare long-term survival of different treatments [22], [30], [32], [36].

In the United States, the positive results of the small GITSG-study and the subsequent supportive study [22], [23] had great impact on therapy tradition. Further use of mostly 5-FU based combination chemoradiotherapy have generated a series of studies, which have been fairly consistent [2], [4], [24–29]. The patients typically show a median survival of 15–26 months, and a 2-year survival of 22–48%. Apart from the initial GITSG-study [22], however, all these studies are non-randomized, and the influence of selection biases is not accounted for. For example, in the large Baltimore experience, decision to deliver therapy was influenced by patient choice, age, slow recovery, and poor performance status [2], [4], [24]. In contrast, the randomized EORTC-study did not find a significant positive effect of chemoradiation [30]. However, the results of this study were weakened, as it included also periampullary cancer, and it was insufficiently powered [31]. The largest randomized study, the ESPAC-1 study, actually showed a negative effect of chemoradiation, when compared with adjuvant chemotherapy or no therapy [36]. Critics have not been lacking, and the 2×2 factorial design, the rate of protocol violations, and the quality assurance and execution of radiotherapy have all been disputed [37], [38], [69], [70].

Adjuvant chemotherapy (without radiotherapy) has been studied in an early randomized study [32], and in the ESPAC-1 study [33–36]. In the Norwegian study (which includes both pancreatic and ampullary cancer), adjuvant chemotherapy resulted in significantly longer median survival, but there was no effect on survival beyond two years [32]. The ESPAC-1 study found a significant positive effect on median survival and 5-year survival [33–36].

A problematic feature of postoperative, adjuvant therapy is that as many as 20–30% of resected patients never receive treatment because of slow recovery or other reasons [2], [22], [23], [30], [50]. Neoadjuvant chemoradiotherapy has been proposed as a more logical way to deliver therapy, as it is supposed to start with the least toxic part of treatment. Thus, proponents of neoadjuvant therapy suggest that the selection process inherent in this modality – in contrast to other adjuvant therapies – can spare surgery in patients, which show metastatic or progressive disease or otherwise prove to be “unfit for surgery”. Further theoretical advantages of preoperative administration have been proposed: (1) well-oxygenated cells may be more radiosensitive than tissue devascularized by surgery; (2) intraoperative spread of cancer cells may be prevented; (3) preoperative treatment may ameliorate concern about narrow or positive resection margins, especially along the superior mesenteric artery; Preoperative treatment seems safe, and does not increase perioperative morbidity or mortality [46–49]. Some studies even suggest, that local postradiation fibrosis may decrease the rate of leakage at the pancreaticojejunal anastomosis [16], [71], [72]. Patients ultimately subjected to pancreatic resection after neoadjuvant therapy have typically shown median survival 19–25 months, 2-year survival 32–61%, and 5-year survival 19–28% (Table III), which compare well with postoperative therapy. However, the selection process introduces a positive bias, and randomized controlled trials showing an overall benefit of neoadjuvant therapy are missing. A further question is to what extent preoperative/neoadjuvant therapy may actually cause down-staging of tumours, as has been claimed in some studies [57–59].

Conclusion

So in 2005, what conclusions can be drawn?

Firstly, randomized controlled trials lend support to the use of adjuvant therapy, and the weight of the large ESPAC-1 study presently suggests the use of 5-FU based chemotherapy. The value of added chemoradiation seems unclear, and needs to be reexamined. However, treatment results with 5-FU based adjuvant therapy are modest, and it is a well-known fact that 5-year survival does not equal cure of pancreatic cancer [8], [12]. 5-FU with leukovorin may not represent the ultimate drug for adjuvant therapy in pancreatic cancer. Gemcitabine was shown to be more effective in advanced pancreatic cancer than short-term infusion of 5-FU alone (although this is an insufficent mode of delivery) [73], [74]. Gemcitabine may be effective also as adjuvant therapy, as indicated by early data from the recent German randomized trial [39]. Further (weak) support is lended by a non-randomized study of 22 patients, where patients receiving gemcitabine and radiotherapy had a median survival of 15 months [75]. Hopefully, this issue can be settled within shortly. Besides the German study, the results of other large randomized controlled studies are awaited. The ESPAC-3 compares adjuvant gemcitabine with adjuvant 5-FU. In the USA, the RTOG 97-04 explores the additional benefit of adjuvant gemcitabine given before and after 5-FU-based chemoradiation [76]. Another strategy would be to further develop multidrug adjuvant treatment, as has been tried also in the past [4], [25], [32], [43], [51], [52], [55], [59]. An interesting example is found in a recent study of 17 patients, which were treated by a combination of 5-FU, cisplatinum, and interferon-α [77]. The regimen resulted in a median survival exceeding 24 months and an estimated 2-year survival of 84%, which represents the highest 2-year survival figure to date [77]. However, further follow-up or extension of this observation has not been reported. Oxaliplatin, which is less toxic than cisplatin, may be more suitable. Preliminary data show a positive effect of oxaliplatin in combination with 5-FU or gemcitabine on advanced pancreatic cancer [78–80].

Secondly, neoadjuvant therapy represents an interesting solution to the problem, although randomized controlled trials are lacking. Complication-prone surgery may be avoided in patients with rapidly progressive disease, and a larger proportion of resected patients may be treated. Although yet unproven, a down-staging effect of preoperative therapy has been claimed [57–59].

Thirdly, locally targeted therapies, i.e. regional chemotherapy and IORT, are difficult to assess, as only small, non-randomized studies exist. Regional chemotherapy appears to result in good survival in the few available studies, whereas intraoperative radiotherapy does not seem to increase long-term survival despite reducing the rate of local recurrence.