| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Editorial |
Department of Surgical Oncology, Fox Chase Cancer Center, 333 Cottman Avenue, Philadelphia, Pennsylvania 19111
Correspondence: Address correspondence and reprint requests to: John Hoffman, MD; E-mail: jp_hoffman{at}fccc.edu
The article by Cheng et al.1 is another attempt to compare the effects of preoperative versus postoperative adjuvant chemoradiotherapy (CRT) for presumably localized pancreatic adenocarcinoma by retrospective analysis of two nonrandomized treatment cohorts.2–4 Because the only significant differences in morbidity within the preoperative and postoperative cohorts in this study were in pancreatic leak and intra-abdominal abscess formation (favoring the preoperative CRT cohort), the primary focus of this study was the relative incidence of postoperative pancreatic leaks. The authors aver that preoperative CRT reduces the incidence of both pancreatic leaks and leak-associated morbidity and mortality, and they even go so far as to recommend preoperative CRT in those with ampullary cancer to decrease postoperative leaks.
Is this information new, true, or relevant? Ishikawa et al.2 were the first to examine the question. They described two retrospective nonrandomized cohorts: one treated with preoperative radiotherapy (RT) and the other not. Patients with fistulas were defined as those with amylase levels >2000 U/L in their drains for >1 week. Furthermore, a fistula was defined as major if either bile was seen in the drain fluid or the fistulogram showed entry of dye into the jejunum (all patients had pancreatojejunostomy over external pancreatic stents). Minor fistulas were those without either of these findings but with the defined drain amylase levels. One of 22 patients with preoperative RT developed a major fistula, and 1 of 54 patients without preoperative RT had a major fistula; 3 of 54 had minor fistulas that became major, and 6 had minor fistulas. Because the two immediate major fistulas were attributed to premature withdrawal or occlusion of pancreatic stents, these cases were excluded. The 9 of 53 patients with minor or minor changing to major fistulas without RT were compared with the 0 of 21 patients without fistulas in the post-RT cohort: differences were significant (P = .05). Statistical significance here, however, required that the two major leaks be omitted. Clearly, those receiving preoperative RT seemed to do no worse than those without it.
Lowy et al.5 examined pancreatic leak in patients with and without preoperative CRT in a randomized study of perioperative octreotide. Their definition of fistula was either biochemical (defined as increased drain amylase at postoperative day 3) or clinical (requiring reoperation or drainage). Whereas none of the 46 patients with protocol-based preoperative CRT for pancreatic cancer developed clinical pancreatic fistulas, there were enough with biochemical leaks that no difference was shown between those treated with preoperative CRT and those who had surgery first. Again, preoperative CRT seemed not to be a deterrent to safe recovery from pancreatoduodenectomy, but leaks were no less frequent than in those treated without preoperative therapy.
Cheng et al.1 defined a leak as any drainage fluid at any time after surgery with a value three times that of serum amylase. Multivariate analysis showed that preoperative CRT produced a statistically significant decrease in pancreatic leaks (10.1% vs. 43.3%; P < .001). For those with pancreatic head cancer, 3 (4.8%) of 62 of those with preoperative CRT had a fistula, compared with 7 (24.1%) of 29 without (P < .006). These are interesting data, and the results support the experience of others that preoperative CRT does not increase perioperative morbidity and mortality, but it does not close the book on this issue.
Investigators for years have attributed decreased postoperative pancreatic leaks to pancreases with severe fibrosis, large ducts, and decreased pancreatic juice production and increased leaks from the soft pancreas with a small, unobstructed duct.2,6–8 Pancreatic acini have been shown to be more radiosensitive than ducts or islets, and RT has been shown to reduce enzyme production.2 However, modern RT for treatment of a tumor in the head of the pancreas (particularly preoperative RT, which usually involves smaller fields) may not treat much of the remaining pancreatic tail. Treatment fields were not described in the article under discussion, so it may be the that delay of surgery by 2 to 3 months, with consequent continued pancreatic ductal dilatation and fibrosis, is more important than any direct effects of CRT on the residual pancreas. If the results of the European Study Group of Pancreatic Cancer 1 trial are corroborated (and, thus, more patients are treated with chemotherapy alone), we may have more data to define the relative effects of time and RT on the pancreatic remnant and its ability to hold sutures.9 However, meta-analysis of all prospective, randomized trials of postoperative adjuvant therapy does show statistically significant improvement in survival of those with R1 (microscopically positive) resection margins treated with CRT, whereas those treated with only chemotherapy did not fare so well.10 Thus, it is likely that CRT will have continued use, particularly for patients with pancreas cancers that are borderline resectable.
Unfortunately, we have no prospective, randomized trial of preoperative versus postoperative adjuvant therapy sequencing. With such a trial, one could follow up two presumably equal cohorts of patients from the time of diagnosis and then be provided with an accurate assessment of attrition and morbidity all the way to recurrence. Without it, we have cohorts of patients who have come to and successfully undergone resection, either with or without prior CRT. Clearly, there is much room for selection bias.
What if the selection process in these retrospective studies favored more advanced lesions going to pre-operative CRT (which is certainly the case at most institutions)? Then the distal glands would tend to be more fibrotic and have more dilated ducts, merely as a function of time, not relating to preoperative CRT per se. What if the surgeons’ techniques varied (one preferring preoperative therapy and another postoperative)? What if the half-life of amylase in the peritoneal cavity is several days, yet some (including these authors) define a leak regardless of the postoperative day? What if RT techniques have changed and continue to change such that the residual pancreas receives much less of a dose? What if there are variables such as splanchnic blood flow and comorbidities that are not yet recognized as important?
Just as important for the question under discussion, we still have no clear consensus on the definition of pancreatic fistula. One suspects that all of the various series may have created their definitions as a service to desired conclusions. The authors have nicely tabulated various definitions but have not subjected their own data and comparisons to these definitions. This calls into question the relevance of this study. Certainly, most patients with mildly increased drain amylase concentrations in the early postoperative period have no significant prolongations of hospital stay. It is the major complication associated with a pancreatic leak that is salient. In this study, major complications associated with leak seem to be more frequent in those without prior CRT, but this is a tenuous conclusion, because the numbers are small and it is often impossible to test for amylase in purulent material.
Would a prospective study of a large group that would correlate increased days of hospitalization and reoperation with drainage amounts and drain amylase production not be a better way to define clinically significant pancreatic leaks? Then this evidence-based definition could be used to more clearly define predisposing factors. Meanwhile, this article by Cheng et al.1 provides further evidence that pancreatic resection succeeding CRT is probably as safe as surgery preceding it. Until either a large meta-analysis of existing preoperative CRT databases or a prospective, randomized trial is performed, one cannot be completely sure that it is either as safe or safer.
Received for publication October 19, 2005. Accepted for publication October 26, 2005.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
