To the Editor:
With great interest we read the article by Pinheiro de Almeida et al.1 on “Transfusion Requirements in Surgical Oncology Patients” that addresses a clinical problem of utmost importance and ongoing debate. However, there remain some concerns with the interpretation of data and conclusions that can be drawn from these results.
Patients were included in this study if their hemoglobin level was more than or equal to 9 g/dl before admission to the intensive care unit (ICU). A hemoglobin level of 9 to 10 g/dl represents a state of severe anemia that occurred in a certain number of patients despite transfusion before randomization (fig. 3). Unfortunately, the incidence and severity of anemia according to the World Health Organization definition in the study groups is not reported. Regarding the further comorbidity profile of patients included, the number of patients with emergency operations (n = 13 vs. 9), congestive heart failure (n = 6 vs. 3), chronic obstructive pulmonary disease (n = 9 vs. 5), diabetes mellitus (n = 26 vs. 20), metastatic cancer disease (n = 39 vs. 32), and cerebrovascular disease (n = 8 vs. 2) were, at least numerically, higher in the restrictive group. It would be interesting whether the composite of these comorbidities was equally distributed between groups to better understand the interaction of anemia and preoperative comorbidities. This would be of great importance for the reader as it has been reported by Musallam et al.2 that the composite postoperative morbidity at 30 days was also higher in patients with anemia than in those without anemia (adjusted odds ratio, 1.35; 95% CI, 1.30 to 1.40). Taken these points together, one may speculate that a higher proportion of sicker patients were randomized into the restrictive group, which could have benefitted from treatment of preoperative anemia to improve their condition before surgery.
Another important limitation that needs to be discussed is that randomization was performed on admission to the ICU. However, during surgery, a mean of 27.8% of patients in the liberal group and 24.8% in the restrictive group have been transfused with one or more units of erythrocytes, resulting in a mean hemoglobin level of 11.0 g/dl in the liberal group and 11.2 g/dl in the restrictive group when being randomized into this study. In this setting, the situation may have occurred that patients were transfused during surgery and being randomized into the restrictive group while patients who were not transfused during surgery could have been randomized to the liberal group having a higher chance of transfusion in the ICU. Both patients may have received the same number of erythrocyte units during surgery and ICU treatment although being in different study groups, which may question the conclusion that a liberal transfusion trigger improves outcome.
The transfusion protocol was followed until discharge from the ICU; however, after discharge from the ICU, transfusion of erythrocytes was left to the discretion of the attending physician until discharge from hospital. The pretransfusion hemoglobin trigger after discharge from the ICU was neither different (mean, 7.5 g/dl in both groups) nor restrictive. This may give rise to the assumption that not a restrictive transfusion practice impaired outcome, but that tolerance of severe anemia with a hemoglobin level between 7 and 9 g/dl during ICU treatment burdens oncologic patients with cardiac, cerebrovascular, and pulmonary comorbidities with a substantial risk of complications (higher incidence of cardiovascular complications with 13.9 vs. 5.2%, P = 0.038, reoperations with 16.8 vs. 10.3%, and abdominal infections with 14.9 vs. 5.2%) and mortality. Preoperative anemia treatment may have prevented hemoglobin levels to decrease until serious adverse effects of anemia become apparent. In addition, the mean ICU stay was 4 days in both groups and survival started to differ between groups after day 12 (fig. 2). This supports the view that severe perioperative anemia may exhibit a delayed negative impact on outcome. In conjunction with the issue of intraoperative transfusions that was described above, this does not necessarily mean that a restrictive transfusion practice impairs complication rate and mortality, but could be attributed to the adverse effects of severe perioperative anemia in oncologic patients.
The pathophysiological explanation for the effect of a more liberal transfusion practice is unclear. The authors speculate that anemia and reduced oxygen delivery may have played a role for impaired tissue oxygenation or microvascular flow. With regard to blood cell transfusion, basic physiology indicates that an increase of hemoglobin enhances oxygen delivery, even though there is still a debate that enhanced oxygen delivery maintains organ function by increasing cellular oxygen uptake and consumption3,4 on a cellular level. In contrast to the results of the presented study, several analyses have shown that the transfusion of erythrocytes has been associated with adverse outcomes, for example, higher infection rates. Furthermore, a large prospective randomized trial in septic shock patients did not describe a higher survival and lower infection rate for a liberal transfusion trigger less than 9 g/dl compared with a more restrictive transfusion at a hemoglobin level less than 7 g/dl.5
In conclusion, to our understanding, it is still controversial whether the restrictive transfusion strategy or the effect of sustained perioperative anemia caused the significant increase in mortality in this group of patients. A prospective trial on the efficacy of a preoperative anemia treatment protocol in patients with cardiac, cerebrovascular, and pulmonary comorbidities undergoing oncologic surgery may answer this question.
Dr. von Heymann has received honoraria for consultancy work and lectures from Vifor Pharma GmbH, Germany, and Janssen-Cilag GmbH, Germany, within the last 36 months. The other authors declare no competing interests.