| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
EDITORIALS |
From the Division of Surgery, City of Hope National Medical Center, Duarte, California.
Correspondence: Address correspondence to: Lawrence D. Wagman, MD, Division of Surgery, City of Hope National Medical Center, 1500 E. Duarte Rd., Duarte, CA 91010; Fax: 626-301-8865; E-mail: lwagman{at}coh.org
Confidence in a clinical observation is strengthened by a large review from a single institution. This is exemplified by the work of Hyung et al.1 reported in this issue, "Adverse Effects of Perioperative Transfusion on Patients with Stage III and IV Gastric Cancer," from the Yonsei University College of Medicine. The 1710 patients they retrospectively reviewed were carefully staged by using strict preoperative and pathologic criteria and were followed with few exclusions (other cancers, gastric remnant cancers, inadequate node dissections, 7.2%) and little loss (3.3%) over an extended period of time (mean follow-up of 70 months). The authors describe for us the details of the transfusion history, dividing the experience into 1 to 2, 3 to 4, and more than 4 U transfused. The narrow time frame of the transfusions includes 14 days preoperatively, intraoperatively, and postoperatively. The mean perioperative hemoglobin levels are reported as 12.1 and 12.9 g/dl, with a statistically significant, but borderline clinical, difference of merely 0.8 g/dl, with the nontransfused group being the higher. Absent in the methodology are the algorithms that were used to trigger a transfusion during the three perioperative periods.
The authors state in the results and discussion that they have identified a dose-response relationship within their bracketing of units transfused (1–2, 3–4, more than 4). This is nicely documented for the entire group of patients. There is little doubt that stage III and IV patients who received a transfusion compared with those who did not had a worse prognosis. However, the dose-dependency data are not included for these two pathologic stages. For practical purposes, when deciding when to transfuse and weighing the risks and benefits of each unit, treating transfusion as a continuous variable, i.e., unit by unit, may be more helpful to the surgeon. This would be particularly true if other strategies were to be used to increase the perioperative hemoglobin and decrease the dose of blood transfused.
The authors, with their data, clearly show poorer survival in transfused stage III and IV gastric cancer patients. Their resultant conclusion from this observation, cause-effect, remains unsubstantiated by use of the data in this paper. However, causality versus association may be the most fascinating and contentious element of the paper. The hypothesis of transfusion-induced immunosuppression, and hence increases in microscopic tumor survival, is not a new one to the oncological literature. The authors cite several papers to support the hypothesis. The theory that the transfusion suppresses the immune response to the patient’s tumor could be tested. Approaches used to support this hypothesis could be to document (1) an increase in patient-specific, gastric-cancer specific antibodies in the nontransfused compared with transfused patients, (2) a relative reduction in these antibodies in patients after transfusion, (3) a relative inability to generate a specific, posttransfusion immune response, or a combination of these. Skilled immunobiologists may be able to attack this challenging problem in a tumor not generally recognized for its immunogenicity.
An interesting comment is that "it is crucial that unnecessary transfusions" be avoided. Considering the known and alleged consequences of transfusion, this statement cannot be challenged. Unless the indications for transfusion are rigorous (e.g., specific hemoglobin level, amount of blood lost in a specific time frame, physiologic parameters linked to blood volume), it is difficult to conclude anything regarding their value or danger. Furthermore, if transfusions are necessary (on the basis of some pre-established criteria, benchmark, or even subtle indications that cannot be clearly defined), then withholding them may have an even more dramatic impact on outcome than their administration. Stimulated by clear cause-effect relationships (e.g., hepatitis c virus and subsequent hepatomas), cost containment, shortages in blood availability, patient safety issues such as transfusion reactions, and medicolegal issues including suits related to transfusion reactions and religious beliefs (e.g., Jehovah’s Witnesses), the medical profession has carefully overseen and regulated the use of blood and blood products. These issues provide a far more powerful argument for reducing unnecessary transfusions through careful patient selection, strict transfusion criteria, and meticulous attention to reducing intraoperative blood loss. The components of transfused blood are complex, and therefore the beneficial and harmful effects of the individual elements must be assessed to properly define the alternatives to blood use, e.g., blood substitutes, hemodilution, erythropoietic therapy, and so on.
Once again the authors have uncovered a critical theme—defining and documenting necessity. An unnecessary transfusion might be defined as one in a patient with a history of bleeding from a gastric cancer with a preoperative hemoglobin of 10 g/dl, no accompanying cardiac problems, and an anticipated blood loss of less than 2 U. This is contrasted with a necessary transfusion, in a patient with history of myocardial ischemia, hemoglobin of 7.0 g/dl, and an expectation of intraoperative blood loss exceeding 1000 ml. It is imperative that the indications for transfusion be linked to the outcome in the transfused patient. Establishing strict criteria for transfusion will allow a far more meaningful analysis of outcome. However, the ultimate conclusion may still be no more definitive than that patients who require transfusion to complete curative operative procedures have shorter survival. The dose-response analysis also takes on far more meaning when the indication for each individual unit is defined and fulfills necessary criteria. Otherwise, dose-response curves may be linked to arbitrary transfusion practices and not to absolute transfusion requirements. In the process of reaching the final and, it is to be hoped, definitive conclusions on the risk of transfusion and the etiology of that risk, the operating surgeon should still adhere to techniques that minimize tissue trauma and blood loss.
Finally, have we allowed the pursuit of prognostic factors to exceed that necessary to make addressable differences in clinical decision-making? At some point, pragmatism must begin to play a role in evaluating the utility of the explosion of indices that have become significant on uni- and multivariant analyses. For a prognostic parameter to be truly valuable, we must use it to either direct or develop therapies. As an example, identification of the presence of tumor in nodes is not important so much because it predicts poorer outcome, but because it allows us to study the effect of adjuvant therapies in a specific risk group. Given the already poor prognosis of stage III and IV gastric cancer patients, would anyone logically withhold therapy because the patient was not transfused or add another agent because a transfusion had been given? Without a cause-effect relationship and a strict algorithm guiding transfusions, they cannot currently be the targets of a therapeutic intervention, i.e., adding a volume expander preoperatively to reduce the number of transfusions or using a blood substitute. We might, however, consider stratifying patients entered into clinical trials on the basis of the presence or absence of transfusion. This would serve to remove what might be a significant confounding prognostic variable for survival.
To conclude, one might make the argument, quite cohesively, that the amount of blood transfused is a collection of subtle measures that are not currently being measured as discrete elements. The challenge before us is to identify what, in this unique or surrogate marker for prognosis, can be measured and manipulated for the good of our patients.
Received for publication October 3, 2001. Accepted for publication October 8, 2001.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
