Adjuvant Therapy for Surgically Resected Pancreatic Adenocarcinoma

In 2008, approximately 37 000 patients will be diagnosed with pancreatic adenocarcinoma in the United States, and most of these patients will succumb to this disease within the first year following diagnosis.¹ Only approximately 7400 patients will have localized cancer, usually involving the head of the pancreas, and are candidates for surgery if the tumor is resectable, as defined by the absence of vascular involvement.² Such patients are candidates for complete resection of the primary cancer. The prognostic significance of an incomplete resection is well-defined (ie, the survival duration of those who undergo an incomplete [R1] resection is on average shorter than that of patients achieving a complete [R0] resection and in some series is no different than the survival of patients with locally advanced stage III disease who receive chemoradiation without surgery, chemotherapy, or both).^{3 - 4} With optimal patient selection, improved surgical techniques, and modern perioperative care, many patients who undergo pancreatic resection will recover adequately to become candidates for postoperative adjuvant therapy and this, coupled with the high frequency of cancer recurrence following surgery, provides a strong impetus to offer such therapy to patients.

The value of therapy following surgical resection of pancreatic adenocarcinoma has been controversial since 1985 following the publication of the first results of the Gastrointestinal Tumor Study Group's (GITSG) underpowered, long-accruing trial.⁵ That study, involving 43 randomized and analyzed patients, demonstrated a survival of 21 months for patients treated with fluorouracil-based chemoradiation and prolonged treatment with systemic fluorouracil compared with an 11-month survival for patients treated with surgery alone.⁵ At that time, the rationale for chemoradiation was that the recurrence pattern of resected pancreatic adenocarcinoma involved local recurrences in the pancreatic bed and metastatic failures involving the liver and peritoneal surfaces necessitating protracted administration of systemic fluorouracil.

Subsequently, 6 additional randomized controlled trials have been published. The most important of these are the European Study Group for Pancreatic Cancer (ESPAC-1) trial⁶ published in 2004 and the Charité Onkologie (CONKO-001) trial⁷ published in 2007. In ESPAC-1, patients who were treated with fluorouracil alone (median survival, 21.6 months) or fluorouracil and fluorouracil-based chemoradiation (median survival, 19.9 months) had improved survival compared with patients treated with surgery alone (median survival, 16.9 months).⁶ In CONKO-001, patients who were treated with postoperative gemcitabine experienced a median disease-free survival of 13.4 months vs 6.9 months for patients treated with surgery alone. This difference was statistically significant (P < .001, log-rank test) yet there was no difference in overall survival between the gemcitabine-treated group (median survival, 22.1 months) and the surgery-alone control group (median survival, 20.2 months).⁷

Now adding to these data are the results of the US Gastrointestinal Intergroup trial RTOG 97-04 led by the Radiation Therapy Oncology Group (RTOG) reported in this issue of JAMA.⁸ Unlike the other trials, this study did not seek to determine whether postoperative therapy was better than surgery alone, but examined whether survival could be extended by substituting gemcitabine for fluorouracil administered systemically before and after fluorouracil-based radiation. Gemcitabine is a nucleoside analog with activity in advanced pancreatic cancer greater than that of fluorouracil.⁹ The strategy of combining both local-regional and systemic therapy persists because despite improvements in preoperative staging, many investigators are still concerned about local as well as distant failure.

Judging from the 34% overall rate of margin-positive resections documented in the RTOG trial, this concern may be well-founded. However, in the ESPAC-1 trial⁶ and the CONKO trial,⁷ the rates of positive resection margins were only 18% and 17%, respectively, and the median survival for the surgery-alone patients in these trials were 16.9 months and 20.1 months, respectively. These results are comparable with the results in RTOG achieved by treatment with postoperative chemotherapy and chemoradiation. These investigators have not developed pathological quality control to the point at which clinicians can be confident that a microscopically positive margin (R1) can be accurately differentiated from a grossly incomplete resection (R2); more attention to the pathological details in future trials is warranted.³ The magnitude of this problem for RTOG is reflected in the 25% of pathology reports that provided no information regarding the status of the surgical margins.

The key result of the RTOG trial⁸ was that for all 451 patients randomized to either fluorouracil or gemcitabine, there was no statistically significant difference in overall or disease-free survival. While not reported in the article, median survival estimates extrapolated from the survival curves (Figure 2 in the article) were approximately 17 months for the fluorouracil group and 19 months for the gemcitabine group. When primary lesions involving the head of the pancreas were considered, there was a nonsignificant increase in median survival for patients treated with gemcitabine vs fluorouracil (20.5 months vs 16.9 months). This finding was further strengthened after adjusting for additional well-established prognostic variables, but did not reach statistical significance.

However, despite the quality control measures taken to ensure state-of-the-art radiation technique in this trial, and even though there was a reduction in local recurrence rates occurring as any component of failure when compared with the GITSG (47%), ESPAC-1 (62% overall), and CONKO-001 (approximately 38% overall) trials, approximately 25% of patients enrolled in the RTOG trial still had some component of local failure. Because local recurrences are usually not isolated, it is uncertain how much these local recurrences detract from the overall survival of these patients.

Because the RTOG trial was not designed to test the benefit of postoperative radiation, the results do little to settle the controversy generated by the reduced survival rates of patients treated with chemoradiation raised by the ESPAC-1 trial—other than to establish that the median survival result among patients with cancer involving the pancreatic head who were treated with gemcitabine in the RTOG trial was comparable with the results of fluorouracil alone and fluorouracil plus fluorouracil-based chemoradiation treatment in the ESPAC-1 trial. In addition, examination of the median survival data from the surgery-alone groups in a contemporary series such as ESPAC-1 and CONKO-001 (16.9 months and 20.1 months, respectively), leads to the provocative suggestion that current postoperative therapeutics may have done much less than expected to improve the survival for patients who have undergone optimal resection. In fact, it is possible that surgical outcomes may be improved further through incorporation of sophisticated diagnostic studies such as pancreas-optimized, dual-phase multidetector computed tomography and endoscopic ultrasonography,¹⁰ coupled with multidisciplinary management and surgery at high-volume centers.

How can outcomes be improved for patients achieving surgical resection (R0) as well as avoiding incomplete resection (≥R1)?⁷ First, the lessons regarding the optimal selection of patients for surgical resection^{2 - 3} need to be exported more effectively into the high-surgical volume setting.¹¹ These criteria can be assessed by the pancreas-optimized, dual-phase multidetector computed tomography and include the absence of extrapancreatic metastases, no evidence of tumor extension to the superior mesenteric artery or celiac axis with a visible tissue plane between the tumor and these structures, and a patent superior mesenteric vein-portal vein confluence. This may have an immediate effect on the survival of patients undergoing pancreatic cancer resections irrespective of postoperative therapy. In addition, the diagnostic and surgical principles that have been refined over the past 25 years need to be incorporated into the design of future prospective clinical trials to ensure that patients entering these trials have a high chance to achieve a R0 resection. Standardized pathological evaluation of the surgical resection margins is critical and needs to be carefully examined and reported.¹²

Second, further work is needed to determine which patients are most likely to benefit from chemoradiation and, if this modality is going to evolve, further emphasis should be placed on the development of more effective radiation sensitizers.¹³

Third, based on interindividual differences in drug metabolism and DNA repair, it appears that patients who are more likely to benefit from current chemotherapy and chemoradiation strategies can be defined prospectively.¹⁴ These individualized approaches should be examined in prospective clinical trials.

Finally, and most importantly, efforts must be redoubled to develop a new generation of promising therapeutics, and the commitment must be increased to understand and test them in prospectively defined patient subsets—not the means to detect marginal or incremental improvements in clinical trials of large numbers of unselected patients.