Transfusion Timing and Postoperative Septic Complications After Gastric Cancer Surgery: Title and subTitle BreakA Retrospective Study of 179 Consecutive Patients FREE

BLOOD TRANSFUSION therapy may result in immunologic changes (immunomodulation) that are beneficial in some patients but harmful in others.^{1 - 4}

The best-characterized clinical effect of transfusion-induced immunomodulation is improved survival of renal allografts in patients who have undergone blood transfusions.⁵ The deleterious effects of transfusion-induced immunosuppression in terms of cancer recurrence and increase in postoperative infectious complications have been debated in the literature for more than a decade, with contradictory results.^{6 - 24} Even if it has been suggested that transfusion may be a marker for other factors that contribute to cancer recurrence, many studies⁴ using multivariate analysis or meta-analysis of individual retrospective studies strongly support the concept that allogenic blood products increase cancer recurrence after a potentially curative surgical resection. The few studies that failed to find significance usually analyzed small populations and were often affected by other methodological problems; however, sometimes no apparent explanation of lack of a transfusion effect can be identified.⁴ The related possibility of an association of perioperative transfusion with postoperative bacterial infection was examined in more than 30 observational studies and in at least 7 randomized, controlled trials. The disagreements between the stated conclusions of the randomized, controlled trials paralleled the discrepant findings of the earlier observational studies.⁶

We carried out a retrospective study to investigate whether perioperative blood transfusions significantly affect postoperative septic morbidity and mortality in patients undergoing elective surgery for gastric cancer. Particular attention was given to evaluation of the relationship between timing of transfusion and postoperative clinical outcome, which has never been focused on in any previous clinical trial.

The hospital records of patients who underwent surgery for gastric cancer at the Department of Surgery, Hospital del Mar, Autonomous University of Barcelona, Barcelona, Spain, and at the Department of Surgery of the Catholic University of Rome, Rome, Italy, from April 1984 to December 1990 were reviewed. We excluded patients with severe organ system derangements such as cardiac disease (documented myocardial infarction within the past 6 months or New York Heart Association class IV or greater), neurologic disease (cerebrovascular accident with persistent neurologic deficit during the past 6 months or neurologic deficit necessitating confinement to a custodial environment at home or in a long-term care facility), renal disease (severe peritoneal dialysis or hemodialysis), hepatic disease (portal hypertension with gastrointestinal tract bleeding or ascites on a hepatocellular basis or hepatic encephalopathy grade II or greater or clinical diagnosis of alcoholic hepatitis), pulmonary disease (long-term supplemental oxygen required at rest), and coagulation defect (prothrombin or partial thromboplastin time ≥2 times the normal value or platelet count of 50 × 10⁹/L or less [≤50000/µL]).

Demographic data, percentage of weight loss, preoperative serum albumin level and lymphocyte count, type and duration of operative procedure, amount and timing of blood transfusion, stage of the disease, hospital morbidity, and mortality were recorded.

Operative procedures consisted of total or subtotal gastrectomy (GR), bypass, or explorative laparotomy.

The total amount of blood transfused was defined as low (<600 mL), intermediate (600-1500 mL), and high (>1500 mL) independently from timing. Patients who did not undergo transfusion were in group A; those who exclusively underwent preoperative and/or perioperative transfusion were in group B; and those who underwent postoperative transfusion, with or without receiving preoperative or perioperative transfusion, were in group C. All patients who underwent transfusion received exclusively whole homologous blood, and in all patients other blood products were not administered during anesthesia or in the preoperative and postoperative periods.

Staging of the disease was performed according to the TNM system.

Postoperative septic complications were classified according to the following criteria: (1) major septic complications: pneumonia, requires radiographic evidence and documentation of a pathologic organism in sputum or pleural fluid; abdominal abscess, requires clinical or instrumental (ultrasound or computed tomographic scan) evidence of an abdominal purulent collection and spontaneous or operative drainage; and septicemia, requires a temperature of 38.5°C or higher and at least 1 blood culture yielding a pathogenic organism; and (2) minor septic complications: wound infection, requires purulent exudate in the wound with or without culture growth; and urinary tract infection, requires bacteriologic confirmation of more than 100000 organisms per 1 mL of urine.

All patients received preoperative short-term antibiotic prophylaxis (1-3 doses of second- or third-generation cephalosporins). No patients underwent chemotherapy or radiotherapy before surgery or within 30 days of hospital discharge.

Statistical analysis was carried out using a software program (Statistical Package for the Social Sciences [SPSS] for Windows, version 6.0, SPSS Inc, Chicago, Ill). The categorical variables were compared using the coded χ² test. P<.05 was considered significant. Univariate analysis was used to evaluate the postoperative septic morbidity with respect to the following variables: age (<70 or ≥70 years), sex, type of treatment (GR, bypass procedures, or explorative laparotomy), duration of surgery (<250 or ≥250 minutes), staging of the disease, nutritional parameters (albumin level, lymphocyte count, and weight loss), need for blood transfusion, and amount and timing of blood transfusion. The same variables were entered into a multivariate logistic regression model.

A total of 209 hospital charts of patients with gastric cancer were reviewed. Of these, 30 patients were excluded because of major concurrent illness, and 179 patients entered the study. Sixty-three patients (35.2%) were women and 116 (64.8%) were men. Mean patient age was 64.6 years (range, 26-92 years). Seventeen patients (9.5%) underwent a bypass procedure, 22 (12.3%) underwent explorative laparotomy, and 140 (78.2%) underwent total or subtotal GR.

The overall septic complication rate was 41.3%. Major septic complications developed in 55 patients (30.7%), 48 of whom had undergone GR. Postoperative mortality was 10.6%; 15 patients (10.7%) in the GR group died and 4 patients (10.3%) in the explorative laparotomy and bypass procedure groups died. Postoperative morbidity and mortality are detailed in Table 1.

One hundred thirty-two patients (73.7%) underwent transfusion. The mean ± SD amount of blood transfused was 1204 ± 1120 mL (range, 300-9600 mL). Fifty-seven patients received a low-volume transfusion, 34 received an intermediate-volume transfusion, and 41 received a high-volume transfusion.

The timing of transfusion is reported in detail in Table 2. There were 47 patients in group A, 91 patients in group B, and 41 patients in group C.

Postoperative mortality and overall major septic complications were significantly higher in patients receiving blood transfusions, but the difference was no longer statistically significant when considering only those undergoing GR (Table 3). Patients who received larger amounts of blood showed a trend toward higher postoperative mortality and septic morbidity (Table 4), particularly concerning development of intra-abdominal abscesses (7.0%, 8.8%, and 21.9% for low, intermediate, and high amounts of blood, respectively; P <.05).

According to the timing of transfusion, postoperative mortality and major septic complication rates were significantly higher in group C (postoperative transfusion) compared with group A (no transfusion) and group B (preoperative or perioperative transfusion) (Table 5). Similar results were obtained considering only patients who underwent GR (Table 5).

Univariate analysis of demographic, nutritional, and treatment-related variables showed a statistically significant association of sex, duration of operative procedure, need for blood transfusion, and amount and timing of transfusion with postoperative major septic complications (Table 6). Of these factors, multivariate logistic regression analysis showed that only lymphocyte count less than 1.5 × 10⁹/L (≤1500/µL) and postoperative administration of blood (group C patients) were independent predictors of septic morbidity (Table 7). Thirty patients (73.2%) in group C developed septic complication. All patients but 1 underwent transfusion after surgery because of postoperative anemia (hemoglobin level, <9 g/L) in the absence of clinical evidence of bleeding at any site. In most of these cases (27 patients), septic complications preceded (19 patients) or were simultaneous with (8 patients) blood transfusion.

Immunosuppression associated with homologous blood transfusion was first observed in renal allograft transplantation.⁵ Clinical evidence of a relationship between transfusion-induced immunosuppression and development of postoperative septic complications has been reported in many series^{6 - 7 ,19 - 24} concerning head and neck, cardiac, orthopedic, and elective and emergency abdominal surgery.

According to these commonly reported figures, results of the present study show a striking relationship between homologous blood transfusion and septic morbidity (35.6% vs 17.0%; P =.01) and mortality in patients who did and did not undergo blood transfusions (13.6% vs 2.1%; P<.05).

In this study, in univariate analysis of factors potentially affecting postoperative septic morbidity, amount and timing of transfusion were statistically significant (Table 6).

A dose-response relationship between transfusion and infection was highlighted by Nichols et al¹⁹ and was evident in the data of Dawes et al²² and in this series. However, this relationship may reflect the effects of an uncontrolled confounder, such as a relationship between hypovolemia and transfusion and between hypovolemia and infections.⁶

Stratifying patients according to timing of transfusion, postoperative mortality, and septic morbidity were similar in groups A and B and significantly lower than in group C (Table 5). Postoperative transfusion was an independent prognostic factor in multivariate analysis (odds ratio, 17.5; 95% confidence interval, 5.8-52.8; P =.01). However, in most group C patients, septic complications preceded or were simultaneous with transfusion, and transfusion was administered in the absence of clinical evidence of bleeding. Therefore, we conclude that extracellular fluid expansion (leading to hemodilution and a low hemoglobin value) during stress response in patients who were developing or had just developed septic complications may act as a confounder and may be considered responsible for the association between postoperative transfusion and septic morbidity. In other words, it is not blood transfusions themselves but the circumstances necessitating transfusions that are the real determinants of prognosis. This does not mean that an immunosuppressive effect of allogenic blood transfusion does not exist but only that it may not be clinically relevant with respect to postoperative infectious complications.

Reprints: Maurizio Bossola, MD, Istituto di Clinica Chirurgica, Università Cattolica del Sacro Cuore, Largo A. Gemelli 8, 00168 Rome, Italy.