EXPERIMENTAL AND CLINICAL TRANSPLANTATION

Objectives: Infection is the most severe complication after an organ transplant. Blood cell transfusion is an independent risk factor for adverse events, including infection in the recipient. This study sought to evaluate the effect of blood product transfusions on nosocomial infections in liver transplant patients.

Materials and Methods: Patients who underwent a liver transplant at our hospital between 2003 and 2010 were recruited for this study. Exclusion criteria were incomplete records, patients who were hospitalized for more than 48 hours during the 4 weeks before transplant, and pediatric transplants. Incidence of nosocomial infections, which were defined as infections occurring within 30 days after transplant was the primary endpoint.

Results: The incidence of nosocomial infections was 28.7%. The number of transfusions of packed red blood cells and fresh frozen plasma was significantly higher in patients with nosocomial infection compared with patients without nosocomial infection (P = .018 and P = .039). Blood products dose-dependently contributed to nosocomial infections. Transfusions of ≥ 7.5 units of red blood cells (odds ratio: 2.8) or ≥ 12.5 units of fresh frozen plasma (odds ratio: 3.27) were associated with nosocomial infections (P = .042 and P = .015). The infection-related mortality rate was 10.3%.

Conclusions: Blood product transfusions are associated with an increased rate of nosocomial infections, which contributes to higher morbidity and mortality.

Key words : Liver transplant, Blood products, Red blood cells, Plasma, Infection

Liver transplant (LT) is the only curative treatment for patients with end-stage liver disease. One of the main factors that increases early posttransplant morbidity and mortality is infection, which affects 8% to 40% of patients,^1-3 with a mortality rate of 15% to 36%.^2,4 Additionally, liver transplant patients are at risk for major blood loss during surgery, because of severe coagulopathies.⁵ Several studies have indicated that allogenic red blood cell (RBC) transfusion is an independent risk factor for adverse events in the recipient; these events include infection, immunologic complications, and multiple organ failure, leading to a higher mortality rate.^6,7 These data apply to trauma, cardiac, and long-term intensive care unit (ICU) patients.^8-11

Although RBC transfusions have been identified as a risk factor for nosocomial infection (NI) after LT, the relation between transfusion of fresh frozen plasma (FFP) and infections in LT patients remains unclear.^12,13 This study sought to evaluate the effect of intraoperatively transfused RBC and FFP on NI, and to determine the cutoff points for each blood product that causes infection in the first month after LT.

Materials and Methods

This study was conducted in accordance with the Helsinki Declaration of 2000. After approval by the Clinical Research Ethics Committee of our institution, the medical records of LT cases were reviewed. Although our LT program was established in 1997, we collected data from patients who were transplanted after 2003, to avoid inaccurate medical records and the adverse effect of the surgical learning curve. We identified 365 adult LT patients in our department from January 2003 to September 2010; of these, 101 consecutive patients (65 men, 36 women) were recruited into the study. Patients who had a hospital stay of more than 48 hours within 1 month before the transplant, pediatric transplants (age < 18 y), and those with incomplete records were excluded from the study.

We recorded the following patient data: age, sex, diagnosis leading to cirrhosis, type of donor, model end-stage liver disease (MELD) score, anesthesia and cold ischemia time, length of stay in the intensive care unit (ICU), use of postoperative mammalian target of rapamycin inhibitors, posttransplant reoperation, retransplant, postoperative renal replacement therapy, mortality rate, type of infection, and numbers of RBC, FFP, and platelet transfusions. The same anesthetic protocol was used in all patients. Postoperatively, all patients were treated at a single ICU with standardized ICU treatment. Perioperative fluid management was designed to maintain a low central venous pressure (5 to 11 mm Hg) by aggressive diuresis and/or nitroglycerine infusion.

Intraoperative transfusion and coagulation management
The transfusion and coagulation therapy followed a strict protocol. Blood products were not given before surgery. During surgery, RBC, or rarely whole blood, was given to maintain a hemoglobin value of 10 g/dL. Intraoperative FFP and platelet transfusions were considered only when there was evidence of decreased clot formation, as determined by conventional coagulation tests. Fresh frozen plasma was transfused when the international normalized ratio was greater than 2 and there was uncontrollable bleeding, with transfusion of 2 or 3 U and hourly checking of international normalized ratio. The threshold for platelet transfusion was regarded as a platelet count < 30 000/μL, regardless of the international normalized ratio. All blood products were prepared according to the recommendations of the American Association of Blood Banks. The recommended storage time for all units of RBCs was 42 days, but they were generally used within 10 to 15 days because of high demand in our hospital. Although all RBC transfusions were prepared with 85% leukoreduction using a high-speed centrifugation method in our center, on occasion we needed to use RBCs without leukoreduction. Plasma was prepared from 80% male and 20% female donors (nonpregnant), and all platelets were allogenic and were leukoreduced on isolation during apheresis.

Perioperative antibiotic prophylaxis and immunosuppressive treatment
Liver transplant recipients received ceftriaxone before induction and postreperfusion as perioperative prophylaxis. Antibiotic prophylaxis was continued during the first 1 to 3 days after transplant. Antiviral and antifungal prophylaxis was not routinely used in the patients.

The immunosuppression protocol consisted of triple immunosuppressive therapy with mycophenolate mofetil, corticosteroids, and cyclosporine or tacrolimus. One gram of methylprednisolone, divided into 2 separate doses, was given before and after graft reperfusion. Starting from postoperative day 1, the daily dosage of methylprednisolone was tapered from 100 mg to 20 mg within 10 days. Corticosteroids and mycophenolate mofetil were given for 6 and 12 months after transplant. Except during the early postoperative period, mammalian target of rapamycin inhibitors (sirolimus or everolimus) were given to patients who experienced adverse effects of the calcineurin inhibitors. Acute graft rejection was treated with 1-gram pulse dosages of methylprednisolone over 2 days.

Definition of posttransplant nosocomial infection
Posttransplant NIs were assessed in all transplant recipients during the first 30 postoperative days, with NI defined according to the Centers for Disease Control and Prevention’s National Nosocomial Infections Surveillance system.¹⁴ Infections occurring ≤ 30 days after an intraoperative transfusion of blood products were considered posttransfusion infections. We assessed pulmonary, blood stream, surgical site, and urinary tract bacterial infections, as well as fungal and cytomegalovirus infections. Any clinically obtained materials from the liver recipients were analyzed and cultured according to routine procedures in a microbiologic laboratory.

Statistical analyses
Data are expressed as means with standard deviations or medians (minimum and maximum values). Differences in age, MELD score, transfused blood products, and presence of infection were evaluated with the Mann-Whiney U test. Sex, donor type, blood type, and MELD score were then evaluated using the chi-square test. Possible cutoff points for RBC and FFP transfusions were estimated by receiver operating characteristic curve analysis. Using the cutoff points, the dose-response relations between transfused blood products and presence of infection were evaluated with the chi-square test for trend. Statistical significance was defined at P < .05.

Of the 365 patients (296 adults and 69 children) who underwent an LT between January 2003 and September 2010 at our hospital, 101 patients (65 men and 36 women) were included in this study, based on our exclusion criteria. The transplanted organs consisted of 89 deceased and 12 living-donor livers. The incidence of NI was 28.7% (29 of 101 patients). Retransplant did not occur in this series. Kidney dysfunction (need for hemodialysis) was not determined in our patients. A posttransplant reoperation was performed in 6.1% of the patients (6/101; 3 each in the infection and noninfection groups). No immunosuppressive treatment was given preoperatively. Mammalian target of rapamycin inhibitors were used in 8.1% of the patients (8/101; 5 in the infection group). Reoperation and mammalian target of rapamycin inhibitor use were not significantly different between the 2 groups (P = .678 and P = .258). The mortality rate was 17.2% in patients with infections (5 patients: 3 had bacterial infections, 1 experienced convulsions, and 1 succumbed to cardiovascular complications). The mortality rate attributed to infection was 10.3% (3 patients).

The effects of age, sex, surgical variables, and transfused blood products on the presence of infection are presented in Tables 1 and 2 . There were no significant differences in age, duration of anesthesia, cold ischemia time, MELD score, platelets, or whole blood between the infection and noninfection groups. However, the median numbers of RBC and FFP units transfused were significantly higher in the infection group. Sex, type of donation (deceased or living), and MELD score ≥ 20 did not affect the incidence of infection.

Incidence and types of infection
Catheter-related and/or bloodstream (n=12, 41.4%) and surgical site infections (n=12, 41.4%) were the most common types of infection. The rates of respiratory, gastrointestinal, and urinary tract infections were 37.9% (n=11), 6.9% (n=2), and 10.3% (n=3). Fifteen of 29 patients (51.7%) had more than 1 type of infection simultaneously. Eleven patients had 2 types of infection, 3 patients had 4, and 3 patients had 5 types of infection. At least 1 bacterial infection was accompanied to all fungal (n=5, 17.2%) and/or cytomegalovirus (n=3, 10.3%) infections. In 16 of the 29 infected patients (55.1%), the pathogen could be determined; the remaining 13 patients had only clinical signs of infection. The most common pathogens were Acinetobacter spp., coagulase-negative staphylococci, Enterococcus spp., and methicillin-resistant Staphylococcus aureus.

Numbers of transfused blood product units and their effect on infection
According to the receiver operating characteristic curve analysis, the area under the curve for RBCs was 65.0%, and the area for FFP was 63.1%. Possible cutoff points (highest sensitivity and lowest 1-specificity) were estimated for RBCs (2.5, 4.5, and 7.5 U) and FFP (6.5 and 12.5 U) (Figure 1). Using these cutoff points, we determined the dose-response relations between blood products and infection (Table 3).

The present study indicates that transfusions of RBCs and FFP are associated with an increased risk for NI. The risk for NI was significantly increased by transfusion of 7.5 U of RBCs or 12.5 U of FFP.

Infection is the most severe complication after solid-organ transplant. Compared with recipients of other solid organs, patients undergoing liver transplant have a higher incidence of infectious complications because of their poor clinical condition (eg, poor nutritional status, comorbidities, exposure to risk factors), the complexity of the surgical procedure, and the use of immunosuppression therapy, which increases susceptibility to opportunistic infections.^15,16 Infections occurring in the early period are most frequently bacterial infections.¹⁷ In the current study, the overall incidence of NI was 28.7%. The most common types of infection were catheter-related and/or bloodstream infections and surgical site infections, with respiratory infections being the next most common infection type.We report a lower incidence of infectious events during the first 30 days (28.7%) as compared with other studies, which report an incidence of 50%¹⁸ and 60%.¹⁹

Vera and associates²⁰ reported that more than one-third of infections (37%) occurred during the first 30 days, and our study agrees with that result. Recently, Benson and associates²¹ reported that 74 of 525 patients (14.1%) developed a postoperative infection, which was associated with an in-hospital mortality of 10.8%. Although their infection rate was lower than ours, their mortality data were consistent with our study, which found an infection-related mortality of 10.3%. Our postoperative care strategies and hospital flora may account in part for the difference in the rate of NI. In the current study, patients developing NI had a more prolonged ICU stay than did patients without infection. Rapid transfer from ICU or direct admission to the ward from the postanesthesia care unit may allow early extubation, withdrawal of indwelling invasive catheters, and mobilization. The different rates of may be due to differences in exclusion criteria among the studies.

We attempted to minimize nontransplant-related risk factors that might have caused postoperative infection by excluding patients who had been hospitalized for 48 hours or longer within 1 month before transplant. In the literature on surgical patients, the relation between the duration of preoperative hospital stay and increased risk for NI is well-defined.^22,23 Bueno Cavanillas and associates²² have demonstrated that the chance for infection is doubled by a preoperative stay of 9 to 20 days, and is increased 5-fold by a stay of more than 20 days; the lowest risk for infection was related to a preoperative stay of 3 to 8 days. Santoro-Lopes and associates²⁴ investigated colonization with methicillin-resistant Staphylococcus aureus infection after LT and found an association between postoperative colonization and hospital admission within 6 months before LT.

During LTs, most patients need major amounts of RBC, FFP, and platelet products. The detrimental effects of intraoperative blood loss and increased perioperative transfusion requirements in LT patients have been assessed in several studies, which showed a correlation between blood use and postoperative morbidity and mortality rates.²⁵ Some studies have reported a high mortality rate secondary to infectious complications in general,^26,27 particularly in patients who had received blood transfusions,²⁸ which is regarded as a strong contributor to morbidity and mortality.²⁹ George and associates³⁰ reported that RBC transfusion of as little as 2 units increased the risk for early abdominal or wound infection. In a study by Cacciarelli and associates,²⁵ all LT patients who received more than 13 U of RBCs developed fatal septic shock. Xia and associates³¹ reported that advanced age, prolonged hospital stay, and greater intraoperative blood transfusion requirements (1400 ± 450 mL) correlated significantly with postoperative pneumonia and mortality. Additionally, Vera and associates²⁰ and Shen and associates³² found that transfusion of > 1000 mL of blood during an operation was an independent risk factor for postoperative infection. Nemes and associates³³ identified risk factors for sepsis after LT in 199 patients. Sepsis was detected in 45 patients (23%) with a mean intraoperative RBC transfusion of 15 ± 8 U, whereas patients without sepsis had received a mean of 10 ± 6.4 U (P = .001). Asensio and associates³ assessed risk factors for surgical site infections, which occurred in 8.8% of their patients as compared with 41.4% of our infected patients. They reported that transfusion of more than 4 U of RBCs was independently associated with the development of SSI; however, in contrast to our results, this correlation was only weakly related to NI in their study. In recent study, Li and associates³⁴ investigated the outcomes of living-donor liver transplant patients who received massive blood transfusion (≥ 6 U RBC). They demonstrated that massive intraoperative blood transfusion can lead to postoperative nosocomial infection.

The association between RBC transfusion and NI has been evaluated according to specific localizations of infections in several studies. A few studies have determined significant cutoff values of RBC transfusion related to postoperative NI.^35,36 Garcia and associates³⁵ reported that the risk of infection increased 1.1-fold with each intraoperative transfusion of 20 mL/kg of packed RBCs, and Patel and associates³⁶ demonstrated that the relative risk for invasive fungal infections was 2.054 when ≥ 22 U of RBCs were transfused. In the present study, the rate of fungal infection was low. The odds ratios of bacterial infection were 2.89 (95% CI, 0.54-15.59) and 4.75 (95% CI, 0.82-27.50) for RBC transfusion of 2.5-4.5 U and 4.5-7.5 U. Compared with less than 2.5 U of RBCs, transfusion of 7.5 or more units increased the infection risk by 2.8 times.

Blood transfusions are especially associated with various immunologic side effects that are believed to result from donor-derived leukocytes, which may have significant immunosuppressive and immunomodulatory effects that are likely to be responsible for the high incidence of postoperative infections.^9,37-40 Therefore, many countries routinely use filters or irradiation to reduce leukocytes in RBCs and concentrates.⁴¹ However, in a preliminary study, Ott and associates⁴² found no benefit from irradiating blood products. In our center, we generally used 85% leukoreduced RBCs, but intraoperative fluids and blood products, including RBCs and FFP, are administered via a Rapid Infusion set (Level 1 System 1000, Smiths Medical, UK) without leukocyte filters.

In the present study, both RBC and FFP transfusions were dose-dependently associated with an increased risk for NI. The risk for bacterial NI was increased 3.27 times greater when ≥ 12.5 U of FFP were transfused. The few previous studies related to FFP transfusions and NI have provided conflicting results. In accord with our results, Patel and associates³⁶ identified an increased FFP transfusion requirement (>16 U) as a risk factor for NI (relative risk, 3.236; 95% CI, 1.770-5.917), including invasive fungal infection. In contrast to our study, Benson and associates²¹ demonstrated that plasma-containing blood products were not a risk factor for developing postoperative NI in liver transplant patients.

As with every clinical study, our study has limitations: The design of our study was retrospective, which carries less persuasive power than prospective randomized trials. We included intraoperative data only, to obtain comparable results. Postoperative transfusions in the ICU, particularly those because of graft failure, were not included. In addition, our sample size was relatively small, and the patients were from only 1 center study, which might have introduced bias. However, the aim of this study was not to determine all of the risk factors for NI, but rather to determine the cutoff points for transfused blood products commonly used in the intraoperative period, to minimize the risk for NI.

Despite these limitations, our study demonstrated that increased transfusions of RBCs and FFP are associated with an increased rate of nosocomial infection. Further studies on a larger scale and including other risk factors for NI are required to elucidate the findings of this research.

Volume: 11 Issue: 6 December 2013

FULL TEXT