Association between red blood cell transfusion and long-term mortality in patients with cancer of the esophagus after esophagectomy

Association between red blood cell transfusion and long-term mortality in patients with cancer of... SUMMARY The impact of red blood cell transfusion on long-term mortality has not been well characterized in patients with cancer of the esophagus after esophagectomy. Our retrospective observational study investigated 611 patients with cancer of the esophagus after esophagectomy from January 2005 to December 2012. Perioperative red blood cell transfusion was defined as red blood cell transfusion during intraoperative and postoperative period. One hundred ninety-six (32.1%) patients received red blood cell transfusion. During follow-up period, 153 (36.9%) patients without red blood cell transfusion and 120 (61.2%) patients with red blood cell transfusion died. Multivariable analysis identified that there was an incremental association between the amount of red blood cell transfusion and long-term mortality (hazard ratio 1.06, 95% confidence interval 1.04–1.08, P < 0.001). The association between red blood cell transfusion and worse long-term mortality was also demonstrated in propensity-matched patients (hazard ratio 1.62, 95% confidence interval 1.15–2.28, P = 0.006). Therefore, there might be an independent association between perioperative red blood cell transfusion and worse long-term mortality in patients with cancer of the esophagus after esophagectomy. Furthermore, there was an incremental increase in long-term mortality in patients who was transfused with red blood cell during perioperative period. INTRODUCTION Red blood cells (RBCs) are transfused to anemic cancer patients to improve oxygen delivery to tissues during perioperative period. However, RBC transfusion is known to be associated with adverse perioperative outcomes, cancer recurrence, and death in different types of cancer patients, including those with bladder, pancreatic, lung, colorectal, and gastric cancer.1-6 It has been speculated that RBC transfusion can induce a systemic inflammatory response and immunomodulation, which can lead to unfavorable outcomes in cancer patients, although the underlying mechanisms remain unclear.7–9 RBC was transfused in up to approximately 60% of patients with cancer of the esophagus who had undergone esophagectomy during perioperative period.10-14 A few studies have addressed the associations between RBC transfusion and long-term mortality in patients with cancer of the esophagus after esophagectomy, but have demonstrated various results. RBC transfusion was found to have detrimental effects on long-term mortality in patients with cancer of the esophagus after esophagectomy in some reports,11-14 whereas other studies reported no effects of RBC transfusion on long-term mortality.10 Furthermore, in these previous reports, RBC transfusion was considered to be a binary variable, and the impacts of RBC transfusion were evaluated in the different two groups based on the number of transfused RBC units.10-14 Hence, even if the association between RBC transfusion and mortality was shown, the previous reports did not identify whether RBC transfusion would have incrementally adverse effects on mortality. The aim of this study is to investigate whether perioperative RBC transfusion would affect long-term mortality in a large number of patients with cancer of the esophagus after esophagectomy. We hypothesized that there would be an incremental association between the amount of RBC transfusion and worse long-term mortality in patients with cancer of the esophagus after esophagectomy. MATERIALS AND METHODS Patients This study included patients with cancer of the esophagus who had undergone esophagectomy at Asan Medical Center (Seoul, Korea), from January 2005 to December 2012. Patients who showed histology other than squamous cell carcinoma and adenocarcinoma had another cancer combined with esophageal cancer, or underwent exploratory thoracotomy were excluded in our analysis. The research protocol was approved by our Institutional Review Board (AMC IRB 2015-0747), and the requirement of written informed consent was waived by AMC IRB. The data were acquired from a retrospective review of electronic medical records. The examined data included age, sex, body mass index (BMI), the year of the surgery, diabetes mellitus on insulin, hypertension, smoking status, history of alcohol consumption, ischemic heart disease, cerebrovascular disease, peripheral vascular disease, chronic obstructive pulmonary disease, chronic kidney disease, history of neoadjuvant chemoradiation therapy, the pretreatment stage of cancer, the American Society of Anesthesiologist (ASA) class, preoperative laboratory data (hemoglobin, creatinine, total bilirubin, and albumin), preoperative left ventricular ejection fraction, preoperative pulmonary function test (predicted forced vital capacity [FVC], predicted forced expiratory volume in one second [FEV1], and FEV1/FVC ratio), intraoperatively infused fluid amount during esophagectomy, and postoperative analgesic strategy (epidural analgesia, intravenous analgesia). The transfused RBC units during intraoperative and postoperative period were examined. The preoperative medications (angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, beta blockers, and statins) were examined. The operation type, minimally invasive operation, resection margin, and postoperative pulmonary complication (pneumonia or acute respiratory distress syndrome) and cardiac complication (acute myocardial infarction, low cardiac output syndrome, ventricular tachycardia or ventricular fibrillation) were also examined. Neoadjuvant chemoradiation therapy has been performed in patients who meet at least one of the following criteria: primary tumor invading the adventitia (T3) or adjacent structure (T4), or evidence of lymph node involvement, or evidence of distant metastasis. The pretreatment stage was categorized into four classes as follows: T1–T2 N0 M0 (stage 0), T3–T4 N0 M0 (stage 1), Tany N+ M0 (stage 2), Tany Nany M1 (stage 3). The cancer's pathologic stage was categorized according to the TNM classification of the 7th ed. of the American Joint Committee on Cancer. Intraoperative and postoperative RBC transfusion management RBC was transfused in patients with hemoglobin of <8 g/dL during intraoperative and postoperative period in patients who underwent esophagectomy at our institution. RBC were also transfused when substantial ongoing bleeding occurred or clinical evidence of end-organ ischemia was observed during intraoperative and postoperative period, even in patients with hemoglobin of 8–10 g/dL, at the discretion of the attending clinicians. In addition, RBC transfusion was considered when patients developed symptoms, including chest pain or tachycardia unresponsiveness to fluid administration, although hemoglobin was 8–10 g/dL during postoperative period. Study endpoints The primary endpoint was the all-cause death after esphagectomy. The cutoff date for death was December 31, 2015. The secondary endpoints were 1-year and 5-year mortality after esophagectomy. STATISTICAL ANALYSIS Continuous data are presented as mean ± standard deviation or median (interquartile range), and the categorical data are presented as frequencies (percentages). We compared the continuous variables with Student t-test or Mann–Whitney U test, and the categorical variables with the χ2-square test or the Fisher's exact test. We used Kaplan–Meier estimates to evaluate long-term mortality, and the difference in survival between patients who received RBC transfusion and those who did not was compared with the log-rank test. To determine the impact of RBC transfusion on long-term mortality, we used three methods to adjust for potential confounding factors. First, univariate and multivariable Cox proportional hazards regression analyses were conducted. All the variables listed in Table 1 were examined, and the variables that showed P values ≤0.10 in the univariate analyses (i.e., age, BMI, diabetes mellitus on insulin, smoking status, peripheral vascular disease, chronic obstructive pulmonary disease, total bilirubin, albumin, neoadjuvant chemoradiation therapy, predicted FEV1, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, beta blockers, crystalloid amount, pathologic stage, and perioperative RBC transfusion) were candidates for multivariable analysis. We used a backward elimination process to construct the final model, and calculated the adjusted hazard ratio (HR) and 95% confidence interval (CI). We undertook sensitivity analyses by performing multivariable analyses, which included all the variables in the primary multivariable model plus the postoperative analgesic strategy or the year of surgery. Perioperative RBC transfusion was considered as both a continuous and categorical variable (no RBC transfusion and RBC transfusion). In addition, we performed multivariable analyses to adjust the pretreatment stage of cancer, operation type, resection margin, and postoperative complication. We confirmed the proportional hazards assumption by checking the log (-log[survival]) curves and by testing the Schoenfeld residuals, and found that there were no relevant violations. Second, we performed propensity score matching analysis. Propensity score were calculated by modeling the probability of receiving RBC transfusion or not receiving RBC transfusion, with the preoperative variables in Table 2. The model showed good calibration (Hosmer–Lemshow test; P = 0.604) and discrimination (c statistic = 0.806). After performing 1:1 to 1:2 propensity score matching, 128 patients who were transfused with RBC were matched to 214 patients who were not. Cox proportional hazards model was employed using propensity score-based matching with robust standard errors. Third, in propensity score matched patients, adjustment was performed through the pathologic stage of cancer, which was not determined before RBC transfusion. Table 1 Baseline characteristics and perioperative data   RBC transfusion  P value    No (n = 415)  Yes (n = 196)    Age (year)  62.0 (57.0–68.0)  64.0 (57.3–69.0)  0.037  Male (%)  394 (94.9)  181 (92.3)  0.204  BMI (kg/m2)  23.6 ± 2.9  22.0 ± 2.8   <0.001  DM on insulin  56 (13.5)  32 (16.3)  0.352  HTN  146 (35.2)  66 (33.7)  0.715  Smoker, current  113 (27.2)  32 (16.3)  0.003  Alc. consumption  366 (88.2)  168 (85.7)  0.389  IHD  8 (1.9)  4 (2.0)  0.999  CVD  12 (2.9)  5 (2.6)  0.811  PVD  11 (2.7)  12 (6.1)  0.035  COPD  3 (7.0)  6 (3.1)  0.034  CKD  15 (3.6)  8 (4.1)  0.777  Neoadjuvant CRT  164 (39.5)  125 (63.8)   <0.001  ASA class      0.621   1  38 (9.2)  14 (7.1)     2  371 (89.4)  178 (90.8)     3  6 (1.4)  4 (2.0)    Hgb (g/dL)  13.1 (12.2–14.0)  11.5 (10.3–13.0)   <0.001  Cr (mg/dL)  0.8 (0.7–1.0)  0.8 (0.7–0.9)   <0.001  T.bil (mg/dL)  0.7 (0.5–0.8)  0.6 (0.5–0.8)  0.002  Alb (mg/dL)  3.9 (3.6–4.1)  3.7 (3.3–3.9)   <0.001  LVEF (%)  62.0 (59.0–65.0)  62.0 (59.0–65.5)  0.305  FVC_p (%)  93.0 (86.0–102.0)  92.0 (82.0–101.0)  0.060  FEV1_p (%)  94.0 (84.5–103.0)  91.0 (82.0–102.0)  0.027  FEV1/FVC (%)  74.0 (68.0–79.0)  74.0 (69.0–80.0)  0.338  ACEI or ARB  67 (16.1)  23 (11.7)  0.151  β-blocker  32 (7.7)  14 (7.1)  0.804  Statin  32 (7.7)  16 (8.2)  0.846  Histologic type      0.394   SCC  407 (98.1)  190 (96.9)     Adeno  8 (1.9)  6 (3.1)    Pretreatment stage†       <0.001   0  250 (60.2)  82 (41.8)     1  34 (8.2)  21 (10.7)     2  117 (28.2)  90 (45.9)     3  8 (1.9)  3 (1.5)     Unknown  6 (1.4)  0 (0)    Crystalloid (intra) (L)  1.3 (0.9–1.9)  1.6 (1.1–2.3)   <0.001  Colloid (intra) (L)  1.0 (0.8–1.2)  1.0 (0.9–1.4)  0.039  Pathologic stage‡      0.005   0  110 (26.5)  53 (27.0)     I  173 (41.7)  56 (28.6)     II  94 (22.7)  50 (25.5)     III  33 (8.0)  32 (16.3)     IV  4 (1.0)  4 (2.0)     Unknown  1 (0.2)  1 (0.5)    Operation type       <0.001   Ivor–Lewis  282 (68.0)  97 (49.5)     Mckeown  115 (27.7)  64 (32.7)     Others$$^\S$$  18 (4.3)  28 (14.3)     Salvage  0 (0)  7 (3.6)    MIO  56 (13.5)  17 (8.7)  0.114  Tumor positive margin  11 (2.7)  16 (8.2)  0.004  Postoperative analgesia      0.681   Epidural  379 (91.3)  177 (90.3)     Intravenous  36 (8.7)  19 (9.7)    Postoperative complication         Pulmonary  36 (8.7)  66 (33.7)   <0.001   Cardiac  53 (12.8)  92 (46.9)   <0.001    RBC transfusion  P value    No (n = 415)  Yes (n = 196)    Age (year)  62.0 (57.0–68.0)  64.0 (57.3–69.0)  0.037  Male (%)  394 (94.9)  181 (92.3)  0.204  BMI (kg/m2)  23.6 ± 2.9  22.0 ± 2.8   <0.001  DM on insulin  56 (13.5)  32 (16.3)  0.352  HTN  146 (35.2)  66 (33.7)  0.715  Smoker, current  113 (27.2)  32 (16.3)  0.003  Alc. consumption  366 (88.2)  168 (85.7)  0.389  IHD  8 (1.9)  4 (2.0)  0.999  CVD  12 (2.9)  5 (2.6)  0.811  PVD  11 (2.7)  12 (6.1)  0.035  COPD  3 (7.0)  6 (3.1)  0.034  CKD  15 (3.6)  8 (4.1)  0.777  Neoadjuvant CRT  164 (39.5)  125 (63.8)   <0.001  ASA class      0.621   1  38 (9.2)  14 (7.1)     2  371 (89.4)  178 (90.8)     3  6 (1.4)  4 (2.0)    Hgb (g/dL)  13.1 (12.2–14.0)  11.5 (10.3–13.0)   <0.001  Cr (mg/dL)  0.8 (0.7–1.0)  0.8 (0.7–0.9)   <0.001  T.bil (mg/dL)  0.7 (0.5–0.8)  0.6 (0.5–0.8)  0.002  Alb (mg/dL)  3.9 (3.6–4.1)  3.7 (3.3–3.9)   <0.001  LVEF (%)  62.0 (59.0–65.0)  62.0 (59.0–65.5)  0.305  FVC_p (%)  93.0 (86.0–102.0)  92.0 (82.0–101.0)  0.060  FEV1_p (%)  94.0 (84.5–103.0)  91.0 (82.0–102.0)  0.027  FEV1/FVC (%)  74.0 (68.0–79.0)  74.0 (69.0–80.0)  0.338  ACEI or ARB  67 (16.1)  23 (11.7)  0.151  β-blocker  32 (7.7)  14 (7.1)  0.804  Statin  32 (7.7)  16 (8.2)  0.846  Histologic type      0.394   SCC  407 (98.1)  190 (96.9)     Adeno  8 (1.9)  6 (3.1)    Pretreatment stage†       <0.001   0  250 (60.2)  82 (41.8)     1  34 (8.2)  21 (10.7)     2  117 (28.2)  90 (45.9)     3  8 (1.9)  3 (1.5)     Unknown  6 (1.4)  0 (0)    Crystalloid (intra) (L)  1.3 (0.9–1.9)  1.6 (1.1–2.3)   <0.001  Colloid (intra) (L)  1.0 (0.8–1.2)  1.0 (0.9–1.4)  0.039  Pathologic stage‡      0.005   0  110 (26.5)  53 (27.0)     I  173 (41.7)  56 (28.6)     II  94 (22.7)  50 (25.5)     III  33 (8.0)  32 (16.3)     IV  4 (1.0)  4 (2.0)     Unknown  1 (0.2)  1 (0.5)    Operation type       <0.001   Ivor–Lewis  282 (68.0)  97 (49.5)     Mckeown  115 (27.7)  64 (32.7)     Others$$^\S$$  18 (4.3)  28 (14.3)     Salvage  0 (0)  7 (3.6)    MIO  56 (13.5)  17 (8.7)  0.114  Tumor positive margin  11 (2.7)  16 (8.2)  0.004  Postoperative analgesia      0.681   Epidural  379 (91.3)  177 (90.3)     Intravenous  36 (8.7)  19 (9.7)    Postoperative complication         Pulmonary  36 (8.7)  66 (33.7)   <0.001   Cardiac  53 (12.8)  92 (46.9)   <0.001  Data are expressed as mean ± standard deviation, median (interquartile range) or number (percentage). †The pretreatment stage 0, 1, 2, and 3 are defined as T1–T2 N0 M0, T3–T4 N0 M0, Tany N+ M0, and Tany Nany M1. ‡The cancer's pathologic stage was categorized according to the TNM classification of the 7th ed. of the American Joint Committee on Cancer. $$^\S$$Others included colon interposition and transhiatal esophagectomy. ACEI or ARB, angiotensin-converting enzyme inhibitor or angiotensin receptor blocker; Adeno, adenocarcinoma; Alb, albumin; Alc, alcohol; ASA, American Society of Anesthesiologist physical status; BMI, body mass index; DM, diabetes mellitus; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; Cr, creatinine; CRT, chemoradiation therapy; CVD, cerebrovascular disease; FEV1_p predicted forced expiratory volume in one second; FVC_p, predicted forced vital capacity; Hgb, hemoglobin; HTN, hypertension; IHD, ischemic heart disease; intra, intraoperative; LVEF, left ventricular ejection fraction; MIO, minimally invasive operation; PVD, peripheral vascular disease; RBC, red blood cell; SCC, squamous cell carcinoma; T.bil, total bilirubin. View Large Table 2 Propensity score-matched data   RBC transfusion  Standardized difference of means    No (n = 214)  Yes (n = 128)    Age (year)  62.1 ± 7.5  62.6 ± 8.0  0.058  Male (%)  198 (92.5)  116 (90.6)  0.068  BMI (kg/m2)  23.0 ± 2.8  22.7 ± 2.6  0.079  DM on insulin  33 (15.4)  18 (14.1)  0.038  HTN  64 (29.9)  38 (29.7)  0.005  Smoker, current  46 (21.5)  26 (20.3)  0.029  Alc. consumption  178 (83.2)  105 (82.0)  0.030  IHD  3 (1.4)  2 (1.6)  0.013  CVD  7 (3.3)  3 (2.3)  0.056  PVD  7 (3.3)  4 (3.1)  0.008  COPD  3 (1.4)  1 (0.8)  0.060  CKD  8 (3.7)  5 (3.9)  0.009  Neoadjuvant CRT  104 (48.6)  67 (52.3)  0.075  ASA class      0.059   1  16 (7.5)  9 (7.0)     2  193 (90.2)  117 (91.4)     3  5 (2.3)  2 (1.6)    Hgb (g/dL)  12.6 ± 1.3  12.4 ± 1.4  0.094  Cr (mg/dL)  0.9 ± 0.2  0.9 ± 0.2  0.043  T.bil (mg/dL)  0.7 ± 0.3  0.7 ± 0.2  0.042  Alb (mg/dL)  3.8 ± 0.4  3.8 ± 0.4  0.023  LVEF (%)  61.7 ± 4.5  61.8 ± 5.2  0.009  FVC_p (%)  92.5 ± 12.4  92.8 ± 13.7  0.020  FEV1_p (%)  92.2 ± 14.6  92.8 ± 18.3  0.035  FEV1/FVC (%)  73.3 ± 9.3  73.6 ± 9.9  0.027  ACEI or ARB  25 (11.7)  13 (10.2)  0.049  β-blocker  13 (6.1)  9 (7.0)  0.039  Statin  14 (6.5)  10 (7.8)  0.049  Histologic type      0.061   SCC  208 (97.2)  123 (96.1)     Adeno  6 (2.8)  5 (3.9)      RBC transfusion  Standardized difference of means    No (n = 214)  Yes (n = 128)    Age (year)  62.1 ± 7.5  62.6 ± 8.0  0.058  Male (%)  198 (92.5)  116 (90.6)  0.068  BMI (kg/m2)  23.0 ± 2.8  22.7 ± 2.6  0.079  DM on insulin  33 (15.4)  18 (14.1)  0.038  HTN  64 (29.9)  38 (29.7)  0.005  Smoker, current  46 (21.5)  26 (20.3)  0.029  Alc. consumption  178 (83.2)  105 (82.0)  0.030  IHD  3 (1.4)  2 (1.6)  0.013  CVD  7 (3.3)  3 (2.3)  0.056  PVD  7 (3.3)  4 (3.1)  0.008  COPD  3 (1.4)  1 (0.8)  0.060  CKD  8 (3.7)  5 (3.9)  0.009  Neoadjuvant CRT  104 (48.6)  67 (52.3)  0.075  ASA class      0.059   1  16 (7.5)  9 (7.0)     2  193 (90.2)  117 (91.4)     3  5 (2.3)  2 (1.6)    Hgb (g/dL)  12.6 ± 1.3  12.4 ± 1.4  0.094  Cr (mg/dL)  0.9 ± 0.2  0.9 ± 0.2  0.043  T.bil (mg/dL)  0.7 ± 0.3  0.7 ± 0.2  0.042  Alb (mg/dL)  3.8 ± 0.4  3.8 ± 0.4  0.023  LVEF (%)  61.7 ± 4.5  61.8 ± 5.2  0.009  FVC_p (%)  92.5 ± 12.4  92.8 ± 13.7  0.020  FEV1_p (%)  92.2 ± 14.6  92.8 ± 18.3  0.035  FEV1/FVC (%)  73.3 ± 9.3  73.6 ± 9.9  0.027  ACEI or ARB  25 (11.7)  13 (10.2)  0.049  β-blocker  13 (6.1)  9 (7.0)  0.039  Statin  14 (6.5)  10 (7.8)  0.049  Histologic type      0.061   SCC  208 (97.2)  123 (96.1)     Adeno  6 (2.8)  5 (3.9)    Data are expressed as mean ± standard deviation or number (percentage). ACEI or ARB, angiotensin-converting enzyme inhibitor or angiotensin receptor blocker; Alb, albumin; Alc, alcohol; Adeno, adenocarcinoma; ASA American Society of Anesthesiologist physical status; BMI, body mass index; CKD chronic kidney disease; CRT, chemoradiation therapy; Cr, creatinine; CVD, cerebrovascular disease; DM, diabetes mellitus; FEV1_p predicted forced expiratory volume in one second; FVC_p predicted forced vital capacity; Hgb, hemoglobin; HTN, hypertension; IHD, ischemic heart disease; LVEF left ventricular ejection fraction; PVD, peripheral vascular disease; RBC, red blood cell, T.bil total bilirubin, SCC, squamous cell carcinoma. View Large A P value <0.05 was considered to be statistically significant. We conducted all the analyses using the SAS software version 9.4 (SAS Institute, Cary, NC). RESULTS Seven hundred patients underwent esophagectomy during our study period. Eighty-nine patients did not satisfy the inclusion criteria as follows; 63 patients had another cancer combined with esophageal cancer, 20 patients showed histology other than squamous cell carcinoma and adenocarcinoma, and six patients underwent exploratory thoracotomy. The remaining 611 patients were evaluated. Patient demographics and perioperative variables are presented in Table 1. A total of 196 (32.1%) patients received RBC transfusion, whereas 415 (67.9%) patients did not. Among the transfused patients, the amount of RBC was 1 unit in 31 (15.8%), 2 units in 79 (40.3%), 3 units in 22 (11.2%), 4 units in 18 (9.2%), and ≥5 units in 46 (23.5%) patients. The patients who received RBC transfusion were older, more likely to have a lower BMI, less likely to be smoker, more likely to have peripheral vascular disease, less likely to have chronic obstructive pulmonary disease, more likely to receive neoadjuvant chemoradiation therapy, more likely to have a lower hemoglobin, creatinine, total bilirubin, and albumin, more likely to show a lower predicted FEV1, more likely to receive a larger amount of intraoperative fluid, and more likely to show a higher pathologic stage of cancer (Table 1). A total of 273 (44.7%) patients died during a median 48.3 months follow up (interquartile range: 25.7–77.4 months). The overall mortality rate was 36.9% in no RBC transfusion group and 61.2% in RBC transfusion group. The 1-year survival rate was 92.8% in no RBC transfusion group and 70.4% in RBC transfusion group, and 5-year survival rate was 62.7% in no RBC transfusion group and 41.0% in RBC transfusion group. The Kaplan–Meier survival curves for RBC transfusion and no RBC transfusion group showed a significant difference (P < 0.001) (Fig. 1). Furthermore, when the survival rates were compared based on the total amount of transfused RBC units, the Kaplan–Meier survival curves revealed different survival rates among patients who received no, 1 unit, 2 units, 3–4 units, and ≥5 units of RBC transfusion (P < 0.001) (Fig. 2). Fig. 1 View largeDownload slide Kaplan–Meier survival curves showed that survival rate was significantly higher in no RBC transfusion group compared to red blood cell transfusion group in the total cohort (A) and in the propensity matched cohort (B) (both P < 0.001). Fig. 1 View largeDownload slide Kaplan–Meier survival curves showed that survival rate was significantly higher in no RBC transfusion group compared to red blood cell transfusion group in the total cohort (A) and in the propensity matched cohort (B) (both P < 0.001). Fig. 2 View largeDownload slide Kaplan–Meier survival curves showed that survival rates were significantly different among patients who received 0, 1, 2, 3-4, and ≥5 units of red blood cell transfusion (P < 0.001). Fig. 2 View largeDownload slide Kaplan–Meier survival curves showed that survival rates were significantly different among patients who received 0, 1, 2, 3-4, and ≥5 units of red blood cell transfusion (P < 0.001). Univariate analysis showed that RBC transfusion was associated with increased long-term mortality (HR 1.05, 95% CI 1.03–1.06, P < 0.001), and multivariable analysis showed that long-term mortality increased by 6.0%, as the amount of transfused RBC increased by 1 unit (HR 1.06, 95% CI 1.04–1.08, P < 0.001) (Table 3). Table 3 Multivariable analyses of factors associated with all-cause long-term mortality in esophageal cancer patients after esophagectomy   Hazard ratio (95% CI)  P value  Age (year)  1.02 (1.01–1.04)  0.004  Body mass index (kg/m2)  0.91 (0.87–0.95)   <0.001  Neoadjuvant CRT  1.69 (1.26–2.27)  0.001  Albumin (mg/dL)  0.55 (0.39–0.79)  0.001  Pathologic stage     <0.001   I  1.46 (0.98–2.18)  0.062   II  3.02 (2.13–4.29)   <0.001   III  5.38 (3.62–7.99)   <0.001   IV  8.60 (3.84–19.27)   <0.001  Red blood cell (units)  1.06 (1.04–1.08)   <0.001    Hazard ratio (95% CI)  P value  Age (year)  1.02 (1.01–1.04)  0.004  Body mass index (kg/m2)  0.91 (0.87–0.95)   <0.001  Neoadjuvant CRT  1.69 (1.26–2.27)  0.001  Albumin (mg/dL)  0.55 (0.39–0.79)  0.001  Pathologic stage     <0.001   I  1.46 (0.98–2.18)  0.062   II  3.02 (2.13–4.29)   <0.001   III  5.38 (3.62–7.99)   <0.001   IV  8.60 (3.84–19.27)   <0.001  Red blood cell (units)  1.06 (1.04–1.08)   <0.001  CI, confidence interval; CRT chemoradiation therapy. View Large Propensity-matched analysis resulted in 128 patients receiving RBC transfusion and 214 patients not receiving RBC transfusion, with similar baseline characteristics (Table 2). The Kaplan–Meier survival curves for RBC transfusion and no RBC transfusion group were significantly different in propensity score-matched patients (P < 0.001) (Fig. 1). Propensity-matched analysis also indicated that RBC transfusion was associated with increased long-term mortality (HR 1.80, 95% CI 1.31–2.48, P < 0.001), and that this association was still significant after adjusting for the pathologic stage of cancer (HR 1.62, 95% CI 1.15–2.28, P = 0.006) (Table 4). Table 4 Association between RBC transfusion and long-term mortality in esophageal cancer patients after esophagectomy   Hazard ratio (95% CI)  P value  RBC transfusion as a continuous variable       Unadjusted  1.05 (1.03–1.06)   <0.001   Multivariable adjusted†  1.06 (1.04–1.08)   <0.001  RBC transfusion as a categorical variable       Unadjusted  2.48 (1.95–3.15)   <0.001   Multivariable adjusted†  1.54 (1.17–2.01)  0.002   Propensity score matching  1.80 (1.31–2.48)   <0.001   Propensity score matching and adjusted by covariate‡  1.62 (1.15–2.28)  0.006    Hazard ratio (95% CI)  P value  RBC transfusion as a continuous variable       Unadjusted  1.05 (1.03–1.06)   <0.001   Multivariable adjusted†  1.06 (1.04–1.08)   <0.001  RBC transfusion as a categorical variable       Unadjusted  2.48 (1.95–3.15)   <0.001   Multivariable adjusted†  1.54 (1.17–2.01)  0.002   Propensity score matching  1.80 (1.31–2.48)   <0.001   Propensity score matching and adjusted by covariate‡  1.62 (1.15–2.28)  0.006  †Adjusted by age, body mass index, neoadjuvant chemoradiation therapy, albumin, and pathologic stage of cancer. ‡Adjustment was performed through pathologic stage of cancer. CI, confidence interval; RBC, red blood cell. View Large In sensitivity analyses, the association between RBC transfusion and long-term mortality remained after adjusting the postoperative analgesic strategy or the year of surgery (Table 5). In addition, the association between RBC transfusion and long-term mortality remained after adjusting the pretreatment stage of cancer or postoperative complications (Table 5). Table 5 Comparison of results for primary analysis compared with sensitivity analyses   Adjusted hazard ratio (95% CI)  P value  RBC transfusion as a continuous variable       Model 1-A†  1.06 (1.04–1.08)   <0.001   Model 1-A + postoperative analgesic strategy  1.06 (1.04–1.08)   <0.001   Model 1-A + the year of surgery  1.07 (1.05–1.08)   <0.001   Model 1-B‡  1.05 (1.03–1.07)   <0.001   Model 1-C$$^\S$$  1.04 (1.03–1.06)   <0.001  RBC transfusion as categorical variable       Model 2-A†  1.54 (1.17–2.01)  0.002   Model 2-A + postoperative analgesic strategy  1.50 (1.14–1.97)  0.004   Model 2-A + the year of surgery  1.69 (1.27–2.24)   <0.001   Model 2-B¶  1.40 (1.06–1.84)  0.017   Model 2-C††  1.54 (1.18–2.03)  0.002    Adjusted hazard ratio (95% CI)  P value  RBC transfusion as a continuous variable       Model 1-A†  1.06 (1.04–1.08)   <0.001   Model 1-A + postoperative analgesic strategy  1.06 (1.04–1.08)   <0.001   Model 1-A + the year of surgery  1.07 (1.05–1.08)   <0.001   Model 1-B‡  1.05 (1.03–1.07)   <0.001   Model 1-C$$^\S$$  1.04 (1.03–1.06)   <0.001  RBC transfusion as categorical variable       Model 2-A†  1.54 (1.17–2.01)  0.002   Model 2-A + postoperative analgesic strategy  1.50 (1.14–1.97)  0.004   Model 2-A + the year of surgery  1.69 (1.27–2.24)   <0.001   Model 2-B¶  1.40 (1.06–1.84)  0.017   Model 2-C††  1.54 (1.18–2.03)  0.002  †Adjusted by age, BMI, neoadjuvant CRT, albumin, and pathologic stage of cancer. ‡Adjusted by age, BMI, neoadjuvant CRT, albumin, pathologic stage of cancer, and postoperative pulmonary complication. $$^\S$$Adjusted by age, BMI, diabetes mellitus on insulin, albumin, beta blockers, pretreatment stage of cancer, and postoperative pulmonary complication. ¶Adjusted by age, BMI, neoadjuvant CRT, albumin, predicted forced expiratory volume in one second, pathologic stage of cancer, and postoperative pulmonary complication. ††Adjusted by age, BMI, diabetes mellitus on insulin, chronic obstructive lung disease, albumin, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, beta blockers, pretreatment stage of cancer, and postoperative pulmonary complication.BMI, body mass index; CI, confidence interval; CRT, chemoradiation therapy; RBC, red blood cell. View Large Eighteen (9.2%) and 0 (0%) patients died during index hospitalization in RBC transfusion group and no RBC transfusion group, respectively, and 21 (10.7%) and 3 (0.7%) died within 90 d after esophagectomy in RBC transfusion group and no RBC transfusion group, respectively. To exclude the effect of RBC transfusion on short-term mortality, we excluded the patients who died during index hospitalization or within 90 d after esophagectomy, and found that association RBC transfusion and long-term mortality still remained (adjusted HR 1.03, 95% CI 1.01–1.06, P = 0.018). DISCUSSION Our present study found that perioperative RBC transfusion was independently associated with worse long-term mortality in a larger number of patients with cancer of the esophagus after esophagectomy. In addition, there was an incremental increase in long-term mortality in those patients who were transfused with RBC during perioperative period. The impacts of RBC transfusion on outcomes have been investigated previously in patients with different types of cancer.1-6 Perioperative RBC transfusion was associated with increased risks of cancer recurrence and death after radical cystectomy in bladder cancer patient,1 and with increased surgical morbidity after colon cancer surgery.2 Associations between perioperative RBC transfusion and adverse outcomes were also observed in patients after gastric, pancreatic, and lung cancer surgery.3-6 Only a few studies have described an association between RBC transfusion and death in patients with cancer of the esophagus, but they reported conflicting results.10-14 One of these studies reported that blood transfusion during perioperative period was associated with worse 1-year survival in stage iii esophageal cancer patients after esophagogastrectomy, but not in patients with esophageal cancer of other stages.10 Moreover, this association was noted only for 1-year survival, but not for 3- to 5-year survival.10 In contrast, other reports indicated that perioperative blood transfusion was associated with worse survival rate in patients with cancer of the esophagus after esophagectomy.11-14 These previous studies demonstrated varying thresholds of RBC units that was related to worse long-term survival. The survival rate decreased in patients who received RBC transfusion equal to or more than the specific threshold, but did not decrease in those who received RBC transfusion less than the specific threshold. The identified thresholds for transfused RBC were varied (2, 3, 4, or 8 units of RBC), among different studies.11-14 This study revealed adverse effects of RBC transfusion on long-term mortality in patients with cancer of the esophagus after esophagectomy, consistent with most previous reports.11-14 Of note, we found that the long-term mortality increased by 6% with a 1 unit increase in transfused RBC, after adjusting for confounders. Thus, our present findings indicate that the long-term mortality gradually increases in patients with cancer of the esophagus, if they received RBC transfusion during perioperative period. Previous studies treated RBC transfusion as a binary variable, and therefore, they did not evaluate whether there was an incremental association between the amount of transfused RBC and mortality. Our present results suggest that perioperative RBC transfusion might have adverse incremental effects on long-term mortality in the 2000s, during which overall survival of esophageal cancer patients improved due to advances in the treatment strategies, such as implementation of neoadjuvant chemoradiation therapy.15 Hence, clinicians need to be cautious when determining RBC transfusion, especially in cases where a large amount of RBC transfusion is required during perioperative period, as the amount of RBC transfusion can affect long-term outcome in esophageal cancer patients after esophagectomy. A potential hypothesis regarding the association between RBC transfusion and worse long-term mortality in cancer patients involves the immunomodulation induced by allogenic blood transfusion, although the underlying mechanisms have not been clearly understood.7-9,16 An association has been found between blood transfusion and gene expression biomarkers related to specific inflammatory pathways.17,18 In trauma patients, the immunosuppressive inflammatory response was reported to be aggravated in transfused patients when compared with those nontransfused; furthermore, blood transfusion was independently associated with immunosuppression after adjusting for injury severity.17 RBC transfusion also activates inflammatory genes in the circulating leukocytes.18 In patients with cancer, the innate immune system is an important contributor which can ameliorates cancer progression.19 In addition, animal study has shown that blood transfusion is an independent predictor for cancer progression.20 We speculate that perioperative RBC transfusion could induce immunosuppression, thus leading to cancer progression, and consequently to an increased risk for death in patients with cancer of the esophagus after esophagectomy. In this study, multivariable analysis indicated that, besides perioperative RBC transfusion, preoperative neoadjuvant chemoradiation therapy was an independent predictor for long-term mortality. We recorded the pathologic stage of cancer that was determined intraoperatively or postoperatively. This pathologic stage does not accurately reflect the preoperative cancer stage that was used to determine whether neoadjuvant chemoradiation therapy should be performed. Therefore, we think neoadjuvant chemoradiation therapy might reflect the severity of esophageal cancer, and was identified to be independent predictor for long-term mortality in patients with cancer of the esophagus. This study has the following limitations to be considered. First, our study was a retrospective observational study, and therefore, it had an inherent limitation imposed by the study design. Unmeasured and unknown confounders could not been adjusted for in statistical analyses, even though we adjusted for several covariates in our model. In addition, we could not completely exclude the possibility that RBC transfusion just indicated more advanced disease or more comorbidities because of our study design. We performed several statistical analyses to adjust various confounders, and we finally found the results were consistent throughout several statistical analyses. We also found that greater exposure to RBC transfusion increased the incidence of the long-term mortality. However, the results need to be cautiously interpreted. Second, it was difficult to identify which one could contribute more than the other between preoperative anemia and perioperative RBC transfusion to postoperative mortality. In our multivariable Cox proportional regression model, preoperative hemoglobin was not adjusted for, because of multicollinearity. However, we calculated the propensity score with the variables obtained prior to RBC transfusion such as the preoperative hemoglobin level. Therefore, our results suggest an independent impact of perioperative RBC transfusion on long-term mortality, irrespective of the preoperative hemoglobin level. In conclusion, we found that perioperative RBC transfusion was associated with worse long-term mortality in patients with cancer of the esophagus after esophagectomy. Especially, the patients who received RBC transfusion showed an incremental increase in mortality. Therefore, the transfusion of RBC during perioperative period may be carefully determined in patients with cancer of the esophagus after esophagectomy, particularly in cases where a large amount of RBC transfusion is required. ACKNOWLEDGMENT The authors would like to thank to Min-Ju Kim, BS, in Department of Clinical Epidemiology and Biostatistics, Asan Medical Center, for providing the professional statistical help. References 1 Linder B J, Frank I, Cheville J C et al.   The impact of perioperative blood transfusion on cancer recurrence and survival following radical cystectomy. Eur Urol  2013; 63: 839– 45. Google Scholar CrossRef Search ADS PubMed  2 Acheson A G, Brookes M J, Spahn D R. Effects of allogeneic red blood cell transfusions on clinical outcomes in patients undergoing colorectal cancer surgery: a systematic review and meta-analysis. Ann Surg  2012; 256: 235– 44. Google Scholar CrossRef Search ADS PubMed  3 Choi J H, Chung H C, Yoo N C et al.   Perioperative blood transfusions and prognosis in patients with curatively resected locally advanced gastric cancer. Oncology  1995; 52: 170– 5. Google Scholar CrossRef Search ADS PubMed  4 Hyung W J, Noh S H, Shin D W et al.   Adverse effects of perioperative transfusion on patients with stage III and IV gastric cancer. Ann Surg Oncol  2002; 9: 5– 12. Google Scholar CrossRef Search ADS PubMed  5 Mavros M N, Xu L, Maqsood H et al.   Perioperative blood transfusion and the prognosis of pancreatic cancer surgery: systematic review and meta-analysis. Ann Surg Oncol  2015; 22: 4382– 91. Google Scholar CrossRef Search ADS PubMed  6 Ng T, Ryder B A, Chern H et al.   Leukocyte-depleted blood transfusion is associated with decreased survival in resected early-stage lung cancer. J Thorac Cardiovasc Surg  2012; 143: 815– 9. Google Scholar CrossRef Search ADS PubMed  7 Vamvakas E C, Blajchman M A. Transfusion-related immunomodulation (TRIM): an update. Blood Rev  2007; 21: 327– 48. Google Scholar CrossRef Search ADS PubMed  8 Cata J P, Wang H, Gottumukkala V, Reuben J, Sessler D I. Inflammatory response, immunosuppression, and cancer recurrence after perioperative blood transfusions. Br J Anaesth  2013; 110: 690– 701. Google Scholar CrossRef Search ADS PubMed  9 Blajchman M A. Immunomodulation and blood transfusion. Am J Ther  2002; 9: 389– 95. Google Scholar CrossRef Search ADS PubMed  10 Craig S R, Adam D J, Yap P L et al.   Effect of blood transfusion on survival after esophagogastrectomy for carcinoma. Ann Thorac Surg  1998; 66: 356– 61. Google Scholar CrossRef Search ADS PubMed  11 Dresner S M, Lamb P J, Shenfine J, Hayes N, Griffin S M. Prognostic significance of perioperative blood transfusion following radical resection for oesophageal carcinoma. Eur J Surg Oncol  2000; 26: 492– 7. Google Scholar CrossRef Search ADS PubMed  12 Swisher S G, Holmes E C, Hunt K K, Gornbein J A, Zinner M J, McFadden DW. Perioperative blood transfusions and decreased long-term survival in esophageal cancer. J Thorac Cardiovasc Surg  1996; 112: 341– 8. Google Scholar CrossRef Search ADS PubMed  13 Tachibana M, Tabara H, Kotoh T et al.   Prognostic significance of perioperative blood transfusions in resectable thoracic esophageal cancer. Am J Gastroenterol  1999; 94: 757– 65. Google Scholar CrossRef Search ADS PubMed  14 Langley S M, Alexiou C, Bailey D H, Weeden D F. The influence of perioperative blood transfusion on survival after esophageal resection for carcinoma. Ann Thorac Surg  2002; 73: 1704– 9. Google Scholar CrossRef Search ADS PubMed  15 Shapiro J, van Lanschot J J, Hulshof M C et al.   Neoadjuvant chemoradiotherapy plus surgery versus surgery alone for oesophageal or junctional cancer (CROSS): long-term results of a randomised controlled trial. Lancet Oncol  2015; 16: 1090– 8. Google Scholar CrossRef Search ADS PubMed  16 Opelz G, Terasaki P I. Improvement of kidney-graft survival with increased numbers of blood transfusions. N Engl J Med  1978; 299: 799– 803. Google Scholar CrossRef Search ADS PubMed  17 Torrance H D, Brohi K, Pearse R M et al.   Association between gene expression biomarkers of immunosuppression and blood transfusion in severely injured polytrauma patients. Ann Surg  2015; 261: 751– 9. Google Scholar CrossRef Search ADS PubMed  18 Escobar G A, Cheng A M, Moore E E et al.   Stored packed red blood cell transfusion up-regulates inflammatory gene expression in circulating leukocytes. Ann Surg  2007; 246: 129– 34. Google Scholar CrossRef Search ADS PubMed  19 Liu Y, Zeng G. Cancer and innate immune system interactions: translational potentials for cancer immunotherapy. J Immunother  2012; 35: 299– 308. Google Scholar CrossRef Search ADS PubMed  20 Atzil S, Arad M, Glasner A et al.   Blood transfusion promotes cancer progression: a critical role for aged erythrocytes. Anesthesiology  2008; 109: 989– 97. Google Scholar CrossRef Search ADS PubMed  © The Authors 2017. Published by Oxford University Press on behalf of International Society for Diseases of the Esophagus. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Diseases of the Esophagus Oxford University Press

Association between red blood cell transfusion and long-term mortality in patients with cancer of the esophagus after esophagectomy

Loading next page...
 
/lp/ou_press/association-between-red-blood-cell-transfusion-and-long-term-mortality-AVCc9mpiok
Publisher
The International Society for Diseases of the Esophagus
Copyright
© The Authors 2017. Published by Oxford University Press on behalf of International Society for Diseases of the Esophagus. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
1120-8694
eISSN
1442-2050
D.O.I.
10.1093/dote/dox123
Publisher site
See Article on Publisher Site

Abstract

SUMMARY The impact of red blood cell transfusion on long-term mortality has not been well characterized in patients with cancer of the esophagus after esophagectomy. Our retrospective observational study investigated 611 patients with cancer of the esophagus after esophagectomy from January 2005 to December 2012. Perioperative red blood cell transfusion was defined as red blood cell transfusion during intraoperative and postoperative period. One hundred ninety-six (32.1%) patients received red blood cell transfusion. During follow-up period, 153 (36.9%) patients without red blood cell transfusion and 120 (61.2%) patients with red blood cell transfusion died. Multivariable analysis identified that there was an incremental association between the amount of red blood cell transfusion and long-term mortality (hazard ratio 1.06, 95% confidence interval 1.04–1.08, P < 0.001). The association between red blood cell transfusion and worse long-term mortality was also demonstrated in propensity-matched patients (hazard ratio 1.62, 95% confidence interval 1.15–2.28, P = 0.006). Therefore, there might be an independent association between perioperative red blood cell transfusion and worse long-term mortality in patients with cancer of the esophagus after esophagectomy. Furthermore, there was an incremental increase in long-term mortality in patients who was transfused with red blood cell during perioperative period. INTRODUCTION Red blood cells (RBCs) are transfused to anemic cancer patients to improve oxygen delivery to tissues during perioperative period. However, RBC transfusion is known to be associated with adverse perioperative outcomes, cancer recurrence, and death in different types of cancer patients, including those with bladder, pancreatic, lung, colorectal, and gastric cancer.1-6 It has been speculated that RBC transfusion can induce a systemic inflammatory response and immunomodulation, which can lead to unfavorable outcomes in cancer patients, although the underlying mechanisms remain unclear.7–9 RBC was transfused in up to approximately 60% of patients with cancer of the esophagus who had undergone esophagectomy during perioperative period.10-14 A few studies have addressed the associations between RBC transfusion and long-term mortality in patients with cancer of the esophagus after esophagectomy, but have demonstrated various results. RBC transfusion was found to have detrimental effects on long-term mortality in patients with cancer of the esophagus after esophagectomy in some reports,11-14 whereas other studies reported no effects of RBC transfusion on long-term mortality.10 Furthermore, in these previous reports, RBC transfusion was considered to be a binary variable, and the impacts of RBC transfusion were evaluated in the different two groups based on the number of transfused RBC units.10-14 Hence, even if the association between RBC transfusion and mortality was shown, the previous reports did not identify whether RBC transfusion would have incrementally adverse effects on mortality. The aim of this study is to investigate whether perioperative RBC transfusion would affect long-term mortality in a large number of patients with cancer of the esophagus after esophagectomy. We hypothesized that there would be an incremental association between the amount of RBC transfusion and worse long-term mortality in patients with cancer of the esophagus after esophagectomy. MATERIALS AND METHODS Patients This study included patients with cancer of the esophagus who had undergone esophagectomy at Asan Medical Center (Seoul, Korea), from January 2005 to December 2012. Patients who showed histology other than squamous cell carcinoma and adenocarcinoma had another cancer combined with esophageal cancer, or underwent exploratory thoracotomy were excluded in our analysis. The research protocol was approved by our Institutional Review Board (AMC IRB 2015-0747), and the requirement of written informed consent was waived by AMC IRB. The data were acquired from a retrospective review of electronic medical records. The examined data included age, sex, body mass index (BMI), the year of the surgery, diabetes mellitus on insulin, hypertension, smoking status, history of alcohol consumption, ischemic heart disease, cerebrovascular disease, peripheral vascular disease, chronic obstructive pulmonary disease, chronic kidney disease, history of neoadjuvant chemoradiation therapy, the pretreatment stage of cancer, the American Society of Anesthesiologist (ASA) class, preoperative laboratory data (hemoglobin, creatinine, total bilirubin, and albumin), preoperative left ventricular ejection fraction, preoperative pulmonary function test (predicted forced vital capacity [FVC], predicted forced expiratory volume in one second [FEV1], and FEV1/FVC ratio), intraoperatively infused fluid amount during esophagectomy, and postoperative analgesic strategy (epidural analgesia, intravenous analgesia). The transfused RBC units during intraoperative and postoperative period were examined. The preoperative medications (angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, beta blockers, and statins) were examined. The operation type, minimally invasive operation, resection margin, and postoperative pulmonary complication (pneumonia or acute respiratory distress syndrome) and cardiac complication (acute myocardial infarction, low cardiac output syndrome, ventricular tachycardia or ventricular fibrillation) were also examined. Neoadjuvant chemoradiation therapy has been performed in patients who meet at least one of the following criteria: primary tumor invading the adventitia (T3) or adjacent structure (T4), or evidence of lymph node involvement, or evidence of distant metastasis. The pretreatment stage was categorized into four classes as follows: T1–T2 N0 M0 (stage 0), T3–T4 N0 M0 (stage 1), Tany N+ M0 (stage 2), Tany Nany M1 (stage 3). The cancer's pathologic stage was categorized according to the TNM classification of the 7th ed. of the American Joint Committee on Cancer. Intraoperative and postoperative RBC transfusion management RBC was transfused in patients with hemoglobin of <8 g/dL during intraoperative and postoperative period in patients who underwent esophagectomy at our institution. RBC were also transfused when substantial ongoing bleeding occurred or clinical evidence of end-organ ischemia was observed during intraoperative and postoperative period, even in patients with hemoglobin of 8–10 g/dL, at the discretion of the attending clinicians. In addition, RBC transfusion was considered when patients developed symptoms, including chest pain or tachycardia unresponsiveness to fluid administration, although hemoglobin was 8–10 g/dL during postoperative period. Study endpoints The primary endpoint was the all-cause death after esphagectomy. The cutoff date for death was December 31, 2015. The secondary endpoints were 1-year and 5-year mortality after esophagectomy. STATISTICAL ANALYSIS Continuous data are presented as mean ± standard deviation or median (interquartile range), and the categorical data are presented as frequencies (percentages). We compared the continuous variables with Student t-test or Mann–Whitney U test, and the categorical variables with the χ2-square test or the Fisher's exact test. We used Kaplan–Meier estimates to evaluate long-term mortality, and the difference in survival between patients who received RBC transfusion and those who did not was compared with the log-rank test. To determine the impact of RBC transfusion on long-term mortality, we used three methods to adjust for potential confounding factors. First, univariate and multivariable Cox proportional hazards regression analyses were conducted. All the variables listed in Table 1 were examined, and the variables that showed P values ≤0.10 in the univariate analyses (i.e., age, BMI, diabetes mellitus on insulin, smoking status, peripheral vascular disease, chronic obstructive pulmonary disease, total bilirubin, albumin, neoadjuvant chemoradiation therapy, predicted FEV1, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, beta blockers, crystalloid amount, pathologic stage, and perioperative RBC transfusion) were candidates for multivariable analysis. We used a backward elimination process to construct the final model, and calculated the adjusted hazard ratio (HR) and 95% confidence interval (CI). We undertook sensitivity analyses by performing multivariable analyses, which included all the variables in the primary multivariable model plus the postoperative analgesic strategy or the year of surgery. Perioperative RBC transfusion was considered as both a continuous and categorical variable (no RBC transfusion and RBC transfusion). In addition, we performed multivariable analyses to adjust the pretreatment stage of cancer, operation type, resection margin, and postoperative complication. We confirmed the proportional hazards assumption by checking the log (-log[survival]) curves and by testing the Schoenfeld residuals, and found that there were no relevant violations. Second, we performed propensity score matching analysis. Propensity score were calculated by modeling the probability of receiving RBC transfusion or not receiving RBC transfusion, with the preoperative variables in Table 2. The model showed good calibration (Hosmer–Lemshow test; P = 0.604) and discrimination (c statistic = 0.806). After performing 1:1 to 1:2 propensity score matching, 128 patients who were transfused with RBC were matched to 214 patients who were not. Cox proportional hazards model was employed using propensity score-based matching with robust standard errors. Third, in propensity score matched patients, adjustment was performed through the pathologic stage of cancer, which was not determined before RBC transfusion. Table 1 Baseline characteristics and perioperative data   RBC transfusion  P value    No (n = 415)  Yes (n = 196)    Age (year)  62.0 (57.0–68.0)  64.0 (57.3–69.0)  0.037  Male (%)  394 (94.9)  181 (92.3)  0.204  BMI (kg/m2)  23.6 ± 2.9  22.0 ± 2.8   <0.001  DM on insulin  56 (13.5)  32 (16.3)  0.352  HTN  146 (35.2)  66 (33.7)  0.715  Smoker, current  113 (27.2)  32 (16.3)  0.003  Alc. consumption  366 (88.2)  168 (85.7)  0.389  IHD  8 (1.9)  4 (2.0)  0.999  CVD  12 (2.9)  5 (2.6)  0.811  PVD  11 (2.7)  12 (6.1)  0.035  COPD  3 (7.0)  6 (3.1)  0.034  CKD  15 (3.6)  8 (4.1)  0.777  Neoadjuvant CRT  164 (39.5)  125 (63.8)   <0.001  ASA class      0.621   1  38 (9.2)  14 (7.1)     2  371 (89.4)  178 (90.8)     3  6 (1.4)  4 (2.0)    Hgb (g/dL)  13.1 (12.2–14.0)  11.5 (10.3–13.0)   <0.001  Cr (mg/dL)  0.8 (0.7–1.0)  0.8 (0.7–0.9)   <0.001  T.bil (mg/dL)  0.7 (0.5–0.8)  0.6 (0.5–0.8)  0.002  Alb (mg/dL)  3.9 (3.6–4.1)  3.7 (3.3–3.9)   <0.001  LVEF (%)  62.0 (59.0–65.0)  62.0 (59.0–65.5)  0.305  FVC_p (%)  93.0 (86.0–102.0)  92.0 (82.0–101.0)  0.060  FEV1_p (%)  94.0 (84.5–103.0)  91.0 (82.0–102.0)  0.027  FEV1/FVC (%)  74.0 (68.0–79.0)  74.0 (69.0–80.0)  0.338  ACEI or ARB  67 (16.1)  23 (11.7)  0.151  β-blocker  32 (7.7)  14 (7.1)  0.804  Statin  32 (7.7)  16 (8.2)  0.846  Histologic type      0.394   SCC  407 (98.1)  190 (96.9)     Adeno  8 (1.9)  6 (3.1)    Pretreatment stage†       <0.001   0  250 (60.2)  82 (41.8)     1  34 (8.2)  21 (10.7)     2  117 (28.2)  90 (45.9)     3  8 (1.9)  3 (1.5)     Unknown  6 (1.4)  0 (0)    Crystalloid (intra) (L)  1.3 (0.9–1.9)  1.6 (1.1–2.3)   <0.001  Colloid (intra) (L)  1.0 (0.8–1.2)  1.0 (0.9–1.4)  0.039  Pathologic stage‡      0.005   0  110 (26.5)  53 (27.0)     I  173 (41.7)  56 (28.6)     II  94 (22.7)  50 (25.5)     III  33 (8.0)  32 (16.3)     IV  4 (1.0)  4 (2.0)     Unknown  1 (0.2)  1 (0.5)    Operation type       <0.001   Ivor–Lewis  282 (68.0)  97 (49.5)     Mckeown  115 (27.7)  64 (32.7)     Others$$^\S$$  18 (4.3)  28 (14.3)     Salvage  0 (0)  7 (3.6)    MIO  56 (13.5)  17 (8.7)  0.114  Tumor positive margin  11 (2.7)  16 (8.2)  0.004  Postoperative analgesia      0.681   Epidural  379 (91.3)  177 (90.3)     Intravenous  36 (8.7)  19 (9.7)    Postoperative complication         Pulmonary  36 (8.7)  66 (33.7)   <0.001   Cardiac  53 (12.8)  92 (46.9)   <0.001    RBC transfusion  P value    No (n = 415)  Yes (n = 196)    Age (year)  62.0 (57.0–68.0)  64.0 (57.3–69.0)  0.037  Male (%)  394 (94.9)  181 (92.3)  0.204  BMI (kg/m2)  23.6 ± 2.9  22.0 ± 2.8   <0.001  DM on insulin  56 (13.5)  32 (16.3)  0.352  HTN  146 (35.2)  66 (33.7)  0.715  Smoker, current  113 (27.2)  32 (16.3)  0.003  Alc. consumption  366 (88.2)  168 (85.7)  0.389  IHD  8 (1.9)  4 (2.0)  0.999  CVD  12 (2.9)  5 (2.6)  0.811  PVD  11 (2.7)  12 (6.1)  0.035  COPD  3 (7.0)  6 (3.1)  0.034  CKD  15 (3.6)  8 (4.1)  0.777  Neoadjuvant CRT  164 (39.5)  125 (63.8)   <0.001  ASA class      0.621   1  38 (9.2)  14 (7.1)     2  371 (89.4)  178 (90.8)     3  6 (1.4)  4 (2.0)    Hgb (g/dL)  13.1 (12.2–14.0)  11.5 (10.3–13.0)   <0.001  Cr (mg/dL)  0.8 (0.7–1.0)  0.8 (0.7–0.9)   <0.001  T.bil (mg/dL)  0.7 (0.5–0.8)  0.6 (0.5–0.8)  0.002  Alb (mg/dL)  3.9 (3.6–4.1)  3.7 (3.3–3.9)   <0.001  LVEF (%)  62.0 (59.0–65.0)  62.0 (59.0–65.5)  0.305  FVC_p (%)  93.0 (86.0–102.0)  92.0 (82.0–101.0)  0.060  FEV1_p (%)  94.0 (84.5–103.0)  91.0 (82.0–102.0)  0.027  FEV1/FVC (%)  74.0 (68.0–79.0)  74.0 (69.0–80.0)  0.338  ACEI or ARB  67 (16.1)  23 (11.7)  0.151  β-blocker  32 (7.7)  14 (7.1)  0.804  Statin  32 (7.7)  16 (8.2)  0.846  Histologic type      0.394   SCC  407 (98.1)  190 (96.9)     Adeno  8 (1.9)  6 (3.1)    Pretreatment stage†       <0.001   0  250 (60.2)  82 (41.8)     1  34 (8.2)  21 (10.7)     2  117 (28.2)  90 (45.9)     3  8 (1.9)  3 (1.5)     Unknown  6 (1.4)  0 (0)    Crystalloid (intra) (L)  1.3 (0.9–1.9)  1.6 (1.1–2.3)   <0.001  Colloid (intra) (L)  1.0 (0.8–1.2)  1.0 (0.9–1.4)  0.039  Pathologic stage‡      0.005   0  110 (26.5)  53 (27.0)     I  173 (41.7)  56 (28.6)     II  94 (22.7)  50 (25.5)     III  33 (8.0)  32 (16.3)     IV  4 (1.0)  4 (2.0)     Unknown  1 (0.2)  1 (0.5)    Operation type       <0.001   Ivor–Lewis  282 (68.0)  97 (49.5)     Mckeown  115 (27.7)  64 (32.7)     Others$$^\S$$  18 (4.3)  28 (14.3)     Salvage  0 (0)  7 (3.6)    MIO  56 (13.5)  17 (8.7)  0.114  Tumor positive margin  11 (2.7)  16 (8.2)  0.004  Postoperative analgesia      0.681   Epidural  379 (91.3)  177 (90.3)     Intravenous  36 (8.7)  19 (9.7)    Postoperative complication         Pulmonary  36 (8.7)  66 (33.7)   <0.001   Cardiac  53 (12.8)  92 (46.9)   <0.001  Data are expressed as mean ± standard deviation, median (interquartile range) or number (percentage). †The pretreatment stage 0, 1, 2, and 3 are defined as T1–T2 N0 M0, T3–T4 N0 M0, Tany N+ M0, and Tany Nany M1. ‡The cancer's pathologic stage was categorized according to the TNM classification of the 7th ed. of the American Joint Committee on Cancer. $$^\S$$Others included colon interposition and transhiatal esophagectomy. ACEI or ARB, angiotensin-converting enzyme inhibitor or angiotensin receptor blocker; Adeno, adenocarcinoma; Alb, albumin; Alc, alcohol; ASA, American Society of Anesthesiologist physical status; BMI, body mass index; DM, diabetes mellitus; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; Cr, creatinine; CRT, chemoradiation therapy; CVD, cerebrovascular disease; FEV1_p predicted forced expiratory volume in one second; FVC_p, predicted forced vital capacity; Hgb, hemoglobin; HTN, hypertension; IHD, ischemic heart disease; intra, intraoperative; LVEF, left ventricular ejection fraction; MIO, minimally invasive operation; PVD, peripheral vascular disease; RBC, red blood cell; SCC, squamous cell carcinoma; T.bil, total bilirubin. View Large Table 2 Propensity score-matched data   RBC transfusion  Standardized difference of means    No (n = 214)  Yes (n = 128)    Age (year)  62.1 ± 7.5  62.6 ± 8.0  0.058  Male (%)  198 (92.5)  116 (90.6)  0.068  BMI (kg/m2)  23.0 ± 2.8  22.7 ± 2.6  0.079  DM on insulin  33 (15.4)  18 (14.1)  0.038  HTN  64 (29.9)  38 (29.7)  0.005  Smoker, current  46 (21.5)  26 (20.3)  0.029  Alc. consumption  178 (83.2)  105 (82.0)  0.030  IHD  3 (1.4)  2 (1.6)  0.013  CVD  7 (3.3)  3 (2.3)  0.056  PVD  7 (3.3)  4 (3.1)  0.008  COPD  3 (1.4)  1 (0.8)  0.060  CKD  8 (3.7)  5 (3.9)  0.009  Neoadjuvant CRT  104 (48.6)  67 (52.3)  0.075  ASA class      0.059   1  16 (7.5)  9 (7.0)     2  193 (90.2)  117 (91.4)     3  5 (2.3)  2 (1.6)    Hgb (g/dL)  12.6 ± 1.3  12.4 ± 1.4  0.094  Cr (mg/dL)  0.9 ± 0.2  0.9 ± 0.2  0.043  T.bil (mg/dL)  0.7 ± 0.3  0.7 ± 0.2  0.042  Alb (mg/dL)  3.8 ± 0.4  3.8 ± 0.4  0.023  LVEF (%)  61.7 ± 4.5  61.8 ± 5.2  0.009  FVC_p (%)  92.5 ± 12.4  92.8 ± 13.7  0.020  FEV1_p (%)  92.2 ± 14.6  92.8 ± 18.3  0.035  FEV1/FVC (%)  73.3 ± 9.3  73.6 ± 9.9  0.027  ACEI or ARB  25 (11.7)  13 (10.2)  0.049  β-blocker  13 (6.1)  9 (7.0)  0.039  Statin  14 (6.5)  10 (7.8)  0.049  Histologic type      0.061   SCC  208 (97.2)  123 (96.1)     Adeno  6 (2.8)  5 (3.9)      RBC transfusion  Standardized difference of means    No (n = 214)  Yes (n = 128)    Age (year)  62.1 ± 7.5  62.6 ± 8.0  0.058  Male (%)  198 (92.5)  116 (90.6)  0.068  BMI (kg/m2)  23.0 ± 2.8  22.7 ± 2.6  0.079  DM on insulin  33 (15.4)  18 (14.1)  0.038  HTN  64 (29.9)  38 (29.7)  0.005  Smoker, current  46 (21.5)  26 (20.3)  0.029  Alc. consumption  178 (83.2)  105 (82.0)  0.030  IHD  3 (1.4)  2 (1.6)  0.013  CVD  7 (3.3)  3 (2.3)  0.056  PVD  7 (3.3)  4 (3.1)  0.008  COPD  3 (1.4)  1 (0.8)  0.060  CKD  8 (3.7)  5 (3.9)  0.009  Neoadjuvant CRT  104 (48.6)  67 (52.3)  0.075  ASA class      0.059   1  16 (7.5)  9 (7.0)     2  193 (90.2)  117 (91.4)     3  5 (2.3)  2 (1.6)    Hgb (g/dL)  12.6 ± 1.3  12.4 ± 1.4  0.094  Cr (mg/dL)  0.9 ± 0.2  0.9 ± 0.2  0.043  T.bil (mg/dL)  0.7 ± 0.3  0.7 ± 0.2  0.042  Alb (mg/dL)  3.8 ± 0.4  3.8 ± 0.4  0.023  LVEF (%)  61.7 ± 4.5  61.8 ± 5.2  0.009  FVC_p (%)  92.5 ± 12.4  92.8 ± 13.7  0.020  FEV1_p (%)  92.2 ± 14.6  92.8 ± 18.3  0.035  FEV1/FVC (%)  73.3 ± 9.3  73.6 ± 9.9  0.027  ACEI or ARB  25 (11.7)  13 (10.2)  0.049  β-blocker  13 (6.1)  9 (7.0)  0.039  Statin  14 (6.5)  10 (7.8)  0.049  Histologic type      0.061   SCC  208 (97.2)  123 (96.1)     Adeno  6 (2.8)  5 (3.9)    Data are expressed as mean ± standard deviation or number (percentage). ACEI or ARB, angiotensin-converting enzyme inhibitor or angiotensin receptor blocker; Alb, albumin; Alc, alcohol; Adeno, adenocarcinoma; ASA American Society of Anesthesiologist physical status; BMI, body mass index; CKD chronic kidney disease; CRT, chemoradiation therapy; Cr, creatinine; CVD, cerebrovascular disease; DM, diabetes mellitus; FEV1_p predicted forced expiratory volume in one second; FVC_p predicted forced vital capacity; Hgb, hemoglobin; HTN, hypertension; IHD, ischemic heart disease; LVEF left ventricular ejection fraction; PVD, peripheral vascular disease; RBC, red blood cell, T.bil total bilirubin, SCC, squamous cell carcinoma. View Large A P value <0.05 was considered to be statistically significant. We conducted all the analyses using the SAS software version 9.4 (SAS Institute, Cary, NC). RESULTS Seven hundred patients underwent esophagectomy during our study period. Eighty-nine patients did not satisfy the inclusion criteria as follows; 63 patients had another cancer combined with esophageal cancer, 20 patients showed histology other than squamous cell carcinoma and adenocarcinoma, and six patients underwent exploratory thoracotomy. The remaining 611 patients were evaluated. Patient demographics and perioperative variables are presented in Table 1. A total of 196 (32.1%) patients received RBC transfusion, whereas 415 (67.9%) patients did not. Among the transfused patients, the amount of RBC was 1 unit in 31 (15.8%), 2 units in 79 (40.3%), 3 units in 22 (11.2%), 4 units in 18 (9.2%), and ≥5 units in 46 (23.5%) patients. The patients who received RBC transfusion were older, more likely to have a lower BMI, less likely to be smoker, more likely to have peripheral vascular disease, less likely to have chronic obstructive pulmonary disease, more likely to receive neoadjuvant chemoradiation therapy, more likely to have a lower hemoglobin, creatinine, total bilirubin, and albumin, more likely to show a lower predicted FEV1, more likely to receive a larger amount of intraoperative fluid, and more likely to show a higher pathologic stage of cancer (Table 1). A total of 273 (44.7%) patients died during a median 48.3 months follow up (interquartile range: 25.7–77.4 months). The overall mortality rate was 36.9% in no RBC transfusion group and 61.2% in RBC transfusion group. The 1-year survival rate was 92.8% in no RBC transfusion group and 70.4% in RBC transfusion group, and 5-year survival rate was 62.7% in no RBC transfusion group and 41.0% in RBC transfusion group. The Kaplan–Meier survival curves for RBC transfusion and no RBC transfusion group showed a significant difference (P < 0.001) (Fig. 1). Furthermore, when the survival rates were compared based on the total amount of transfused RBC units, the Kaplan–Meier survival curves revealed different survival rates among patients who received no, 1 unit, 2 units, 3–4 units, and ≥5 units of RBC transfusion (P < 0.001) (Fig. 2). Fig. 1 View largeDownload slide Kaplan–Meier survival curves showed that survival rate was significantly higher in no RBC transfusion group compared to red blood cell transfusion group in the total cohort (A) and in the propensity matched cohort (B) (both P < 0.001). Fig. 1 View largeDownload slide Kaplan–Meier survival curves showed that survival rate was significantly higher in no RBC transfusion group compared to red blood cell transfusion group in the total cohort (A) and in the propensity matched cohort (B) (both P < 0.001). Fig. 2 View largeDownload slide Kaplan–Meier survival curves showed that survival rates were significantly different among patients who received 0, 1, 2, 3-4, and ≥5 units of red blood cell transfusion (P < 0.001). Fig. 2 View largeDownload slide Kaplan–Meier survival curves showed that survival rates were significantly different among patients who received 0, 1, 2, 3-4, and ≥5 units of red blood cell transfusion (P < 0.001). Univariate analysis showed that RBC transfusion was associated with increased long-term mortality (HR 1.05, 95% CI 1.03–1.06, P < 0.001), and multivariable analysis showed that long-term mortality increased by 6.0%, as the amount of transfused RBC increased by 1 unit (HR 1.06, 95% CI 1.04–1.08, P < 0.001) (Table 3). Table 3 Multivariable analyses of factors associated with all-cause long-term mortality in esophageal cancer patients after esophagectomy   Hazard ratio (95% CI)  P value  Age (year)  1.02 (1.01–1.04)  0.004  Body mass index (kg/m2)  0.91 (0.87–0.95)   <0.001  Neoadjuvant CRT  1.69 (1.26–2.27)  0.001  Albumin (mg/dL)  0.55 (0.39–0.79)  0.001  Pathologic stage     <0.001   I  1.46 (0.98–2.18)  0.062   II  3.02 (2.13–4.29)   <0.001   III  5.38 (3.62–7.99)   <0.001   IV  8.60 (3.84–19.27)   <0.001  Red blood cell (units)  1.06 (1.04–1.08)   <0.001    Hazard ratio (95% CI)  P value  Age (year)  1.02 (1.01–1.04)  0.004  Body mass index (kg/m2)  0.91 (0.87–0.95)   <0.001  Neoadjuvant CRT  1.69 (1.26–2.27)  0.001  Albumin (mg/dL)  0.55 (0.39–0.79)  0.001  Pathologic stage     <0.001   I  1.46 (0.98–2.18)  0.062   II  3.02 (2.13–4.29)   <0.001   III  5.38 (3.62–7.99)   <0.001   IV  8.60 (3.84–19.27)   <0.001  Red blood cell (units)  1.06 (1.04–1.08)   <0.001  CI, confidence interval; CRT chemoradiation therapy. View Large Propensity-matched analysis resulted in 128 patients receiving RBC transfusion and 214 patients not receiving RBC transfusion, with similar baseline characteristics (Table 2). The Kaplan–Meier survival curves for RBC transfusion and no RBC transfusion group were significantly different in propensity score-matched patients (P < 0.001) (Fig. 1). Propensity-matched analysis also indicated that RBC transfusion was associated with increased long-term mortality (HR 1.80, 95% CI 1.31–2.48, P < 0.001), and that this association was still significant after adjusting for the pathologic stage of cancer (HR 1.62, 95% CI 1.15–2.28, P = 0.006) (Table 4). Table 4 Association between RBC transfusion and long-term mortality in esophageal cancer patients after esophagectomy   Hazard ratio (95% CI)  P value  RBC transfusion as a continuous variable       Unadjusted  1.05 (1.03–1.06)   <0.001   Multivariable adjusted†  1.06 (1.04–1.08)   <0.001  RBC transfusion as a categorical variable       Unadjusted  2.48 (1.95–3.15)   <0.001   Multivariable adjusted†  1.54 (1.17–2.01)  0.002   Propensity score matching  1.80 (1.31–2.48)   <0.001   Propensity score matching and adjusted by covariate‡  1.62 (1.15–2.28)  0.006    Hazard ratio (95% CI)  P value  RBC transfusion as a continuous variable       Unadjusted  1.05 (1.03–1.06)   <0.001   Multivariable adjusted†  1.06 (1.04–1.08)   <0.001  RBC transfusion as a categorical variable       Unadjusted  2.48 (1.95–3.15)   <0.001   Multivariable adjusted†  1.54 (1.17–2.01)  0.002   Propensity score matching  1.80 (1.31–2.48)   <0.001   Propensity score matching and adjusted by covariate‡  1.62 (1.15–2.28)  0.006  †Adjusted by age, body mass index, neoadjuvant chemoradiation therapy, albumin, and pathologic stage of cancer. ‡Adjustment was performed through pathologic stage of cancer. CI, confidence interval; RBC, red blood cell. View Large In sensitivity analyses, the association between RBC transfusion and long-term mortality remained after adjusting the postoperative analgesic strategy or the year of surgery (Table 5). In addition, the association between RBC transfusion and long-term mortality remained after adjusting the pretreatment stage of cancer or postoperative complications (Table 5). Table 5 Comparison of results for primary analysis compared with sensitivity analyses   Adjusted hazard ratio (95% CI)  P value  RBC transfusion as a continuous variable       Model 1-A†  1.06 (1.04–1.08)   <0.001   Model 1-A + postoperative analgesic strategy  1.06 (1.04–1.08)   <0.001   Model 1-A + the year of surgery  1.07 (1.05–1.08)   <0.001   Model 1-B‡  1.05 (1.03–1.07)   <0.001   Model 1-C$$^\S$$  1.04 (1.03–1.06)   <0.001  RBC transfusion as categorical variable       Model 2-A†  1.54 (1.17–2.01)  0.002   Model 2-A + postoperative analgesic strategy  1.50 (1.14–1.97)  0.004   Model 2-A + the year of surgery  1.69 (1.27–2.24)   <0.001   Model 2-B¶  1.40 (1.06–1.84)  0.017   Model 2-C††  1.54 (1.18–2.03)  0.002    Adjusted hazard ratio (95% CI)  P value  RBC transfusion as a continuous variable       Model 1-A†  1.06 (1.04–1.08)   <0.001   Model 1-A + postoperative analgesic strategy  1.06 (1.04–1.08)   <0.001   Model 1-A + the year of surgery  1.07 (1.05–1.08)   <0.001   Model 1-B‡  1.05 (1.03–1.07)   <0.001   Model 1-C$$^\S$$  1.04 (1.03–1.06)   <0.001  RBC transfusion as categorical variable       Model 2-A†  1.54 (1.17–2.01)  0.002   Model 2-A + postoperative analgesic strategy  1.50 (1.14–1.97)  0.004   Model 2-A + the year of surgery  1.69 (1.27–2.24)   <0.001   Model 2-B¶  1.40 (1.06–1.84)  0.017   Model 2-C††  1.54 (1.18–2.03)  0.002  †Adjusted by age, BMI, neoadjuvant CRT, albumin, and pathologic stage of cancer. ‡Adjusted by age, BMI, neoadjuvant CRT, albumin, pathologic stage of cancer, and postoperative pulmonary complication. $$^\S$$Adjusted by age, BMI, diabetes mellitus on insulin, albumin, beta blockers, pretreatment stage of cancer, and postoperative pulmonary complication. ¶Adjusted by age, BMI, neoadjuvant CRT, albumin, predicted forced expiratory volume in one second, pathologic stage of cancer, and postoperative pulmonary complication. ††Adjusted by age, BMI, diabetes mellitus on insulin, chronic obstructive lung disease, albumin, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, beta blockers, pretreatment stage of cancer, and postoperative pulmonary complication.BMI, body mass index; CI, confidence interval; CRT, chemoradiation therapy; RBC, red blood cell. View Large Eighteen (9.2%) and 0 (0%) patients died during index hospitalization in RBC transfusion group and no RBC transfusion group, respectively, and 21 (10.7%) and 3 (0.7%) died within 90 d after esophagectomy in RBC transfusion group and no RBC transfusion group, respectively. To exclude the effect of RBC transfusion on short-term mortality, we excluded the patients who died during index hospitalization or within 90 d after esophagectomy, and found that association RBC transfusion and long-term mortality still remained (adjusted HR 1.03, 95% CI 1.01–1.06, P = 0.018). DISCUSSION Our present study found that perioperative RBC transfusion was independently associated with worse long-term mortality in a larger number of patients with cancer of the esophagus after esophagectomy. In addition, there was an incremental increase in long-term mortality in those patients who were transfused with RBC during perioperative period. The impacts of RBC transfusion on outcomes have been investigated previously in patients with different types of cancer.1-6 Perioperative RBC transfusion was associated with increased risks of cancer recurrence and death after radical cystectomy in bladder cancer patient,1 and with increased surgical morbidity after colon cancer surgery.2 Associations between perioperative RBC transfusion and adverse outcomes were also observed in patients after gastric, pancreatic, and lung cancer surgery.3-6 Only a few studies have described an association between RBC transfusion and death in patients with cancer of the esophagus, but they reported conflicting results.10-14 One of these studies reported that blood transfusion during perioperative period was associated with worse 1-year survival in stage iii esophageal cancer patients after esophagogastrectomy, but not in patients with esophageal cancer of other stages.10 Moreover, this association was noted only for 1-year survival, but not for 3- to 5-year survival.10 In contrast, other reports indicated that perioperative blood transfusion was associated with worse survival rate in patients with cancer of the esophagus after esophagectomy.11-14 These previous studies demonstrated varying thresholds of RBC units that was related to worse long-term survival. The survival rate decreased in patients who received RBC transfusion equal to or more than the specific threshold, but did not decrease in those who received RBC transfusion less than the specific threshold. The identified thresholds for transfused RBC were varied (2, 3, 4, or 8 units of RBC), among different studies.11-14 This study revealed adverse effects of RBC transfusion on long-term mortality in patients with cancer of the esophagus after esophagectomy, consistent with most previous reports.11-14 Of note, we found that the long-term mortality increased by 6% with a 1 unit increase in transfused RBC, after adjusting for confounders. Thus, our present findings indicate that the long-term mortality gradually increases in patients with cancer of the esophagus, if they received RBC transfusion during perioperative period. Previous studies treated RBC transfusion as a binary variable, and therefore, they did not evaluate whether there was an incremental association between the amount of transfused RBC and mortality. Our present results suggest that perioperative RBC transfusion might have adverse incremental effects on long-term mortality in the 2000s, during which overall survival of esophageal cancer patients improved due to advances in the treatment strategies, such as implementation of neoadjuvant chemoradiation therapy.15 Hence, clinicians need to be cautious when determining RBC transfusion, especially in cases where a large amount of RBC transfusion is required during perioperative period, as the amount of RBC transfusion can affect long-term outcome in esophageal cancer patients after esophagectomy. A potential hypothesis regarding the association between RBC transfusion and worse long-term mortality in cancer patients involves the immunomodulation induced by allogenic blood transfusion, although the underlying mechanisms have not been clearly understood.7-9,16 An association has been found between blood transfusion and gene expression biomarkers related to specific inflammatory pathways.17,18 In trauma patients, the immunosuppressive inflammatory response was reported to be aggravated in transfused patients when compared with those nontransfused; furthermore, blood transfusion was independently associated with immunosuppression after adjusting for injury severity.17 RBC transfusion also activates inflammatory genes in the circulating leukocytes.18 In patients with cancer, the innate immune system is an important contributor which can ameliorates cancer progression.19 In addition, animal study has shown that blood transfusion is an independent predictor for cancer progression.20 We speculate that perioperative RBC transfusion could induce immunosuppression, thus leading to cancer progression, and consequently to an increased risk for death in patients with cancer of the esophagus after esophagectomy. In this study, multivariable analysis indicated that, besides perioperative RBC transfusion, preoperative neoadjuvant chemoradiation therapy was an independent predictor for long-term mortality. We recorded the pathologic stage of cancer that was determined intraoperatively or postoperatively. This pathologic stage does not accurately reflect the preoperative cancer stage that was used to determine whether neoadjuvant chemoradiation therapy should be performed. Therefore, we think neoadjuvant chemoradiation therapy might reflect the severity of esophageal cancer, and was identified to be independent predictor for long-term mortality in patients with cancer of the esophagus. This study has the following limitations to be considered. First, our study was a retrospective observational study, and therefore, it had an inherent limitation imposed by the study design. Unmeasured and unknown confounders could not been adjusted for in statistical analyses, even though we adjusted for several covariates in our model. In addition, we could not completely exclude the possibility that RBC transfusion just indicated more advanced disease or more comorbidities because of our study design. We performed several statistical analyses to adjust various confounders, and we finally found the results were consistent throughout several statistical analyses. We also found that greater exposure to RBC transfusion increased the incidence of the long-term mortality. However, the results need to be cautiously interpreted. Second, it was difficult to identify which one could contribute more than the other between preoperative anemia and perioperative RBC transfusion to postoperative mortality. In our multivariable Cox proportional regression model, preoperative hemoglobin was not adjusted for, because of multicollinearity. However, we calculated the propensity score with the variables obtained prior to RBC transfusion such as the preoperative hemoglobin level. Therefore, our results suggest an independent impact of perioperative RBC transfusion on long-term mortality, irrespective of the preoperative hemoglobin level. In conclusion, we found that perioperative RBC transfusion was associated with worse long-term mortality in patients with cancer of the esophagus after esophagectomy. Especially, the patients who received RBC transfusion showed an incremental increase in mortality. Therefore, the transfusion of RBC during perioperative period may be carefully determined in patients with cancer of the esophagus after esophagectomy, particularly in cases where a large amount of RBC transfusion is required. ACKNOWLEDGMENT The authors would like to thank to Min-Ju Kim, BS, in Department of Clinical Epidemiology and Biostatistics, Asan Medical Center, for providing the professional statistical help. References 1 Linder B J, Frank I, Cheville J C et al.   The impact of perioperative blood transfusion on cancer recurrence and survival following radical cystectomy. Eur Urol  2013; 63: 839– 45. Google Scholar CrossRef Search ADS PubMed  2 Acheson A G, Brookes M J, Spahn D R. Effects of allogeneic red blood cell transfusions on clinical outcomes in patients undergoing colorectal cancer surgery: a systematic review and meta-analysis. Ann Surg  2012; 256: 235– 44. Google Scholar CrossRef Search ADS PubMed  3 Choi J H, Chung H C, Yoo N C et al.   Perioperative blood transfusions and prognosis in patients with curatively resected locally advanced gastric cancer. Oncology  1995; 52: 170– 5. Google Scholar CrossRef Search ADS PubMed  4 Hyung W J, Noh S H, Shin D W et al.   Adverse effects of perioperative transfusion on patients with stage III and IV gastric cancer. Ann Surg Oncol  2002; 9: 5– 12. Google Scholar CrossRef Search ADS PubMed  5 Mavros M N, Xu L, Maqsood H et al.   Perioperative blood transfusion and the prognosis of pancreatic cancer surgery: systematic review and meta-analysis. Ann Surg Oncol  2015; 22: 4382– 91. Google Scholar CrossRef Search ADS PubMed  6 Ng T, Ryder B A, Chern H et al.   Leukocyte-depleted blood transfusion is associated with decreased survival in resected early-stage lung cancer. J Thorac Cardiovasc Surg  2012; 143: 815– 9. Google Scholar CrossRef Search ADS PubMed  7 Vamvakas E C, Blajchman M A. Transfusion-related immunomodulation (TRIM): an update. Blood Rev  2007; 21: 327– 48. Google Scholar CrossRef Search ADS PubMed  8 Cata J P, Wang H, Gottumukkala V, Reuben J, Sessler D I. Inflammatory response, immunosuppression, and cancer recurrence after perioperative blood transfusions. Br J Anaesth  2013; 110: 690– 701. Google Scholar CrossRef Search ADS PubMed  9 Blajchman M A. Immunomodulation and blood transfusion. Am J Ther  2002; 9: 389– 95. Google Scholar CrossRef Search ADS PubMed  10 Craig S R, Adam D J, Yap P L et al.   Effect of blood transfusion on survival after esophagogastrectomy for carcinoma. Ann Thorac Surg  1998; 66: 356– 61. Google Scholar CrossRef Search ADS PubMed  11 Dresner S M, Lamb P J, Shenfine J, Hayes N, Griffin S M. Prognostic significance of perioperative blood transfusion following radical resection for oesophageal carcinoma. Eur J Surg Oncol  2000; 26: 492– 7. Google Scholar CrossRef Search ADS PubMed  12 Swisher S G, Holmes E C, Hunt K K, Gornbein J A, Zinner M J, McFadden DW. Perioperative blood transfusions and decreased long-term survival in esophageal cancer. J Thorac Cardiovasc Surg  1996; 112: 341– 8. Google Scholar CrossRef Search ADS PubMed  13 Tachibana M, Tabara H, Kotoh T et al.   Prognostic significance of perioperative blood transfusions in resectable thoracic esophageal cancer. Am J Gastroenterol  1999; 94: 757– 65. Google Scholar CrossRef Search ADS PubMed  14 Langley S M, Alexiou C, Bailey D H, Weeden D F. The influence of perioperative blood transfusion on survival after esophageal resection for carcinoma. Ann Thorac Surg  2002; 73: 1704– 9. Google Scholar CrossRef Search ADS PubMed  15 Shapiro J, van Lanschot J J, Hulshof M C et al.   Neoadjuvant chemoradiotherapy plus surgery versus surgery alone for oesophageal or junctional cancer (CROSS): long-term results of a randomised controlled trial. Lancet Oncol  2015; 16: 1090– 8. Google Scholar CrossRef Search ADS PubMed  16 Opelz G, Terasaki P I. Improvement of kidney-graft survival with increased numbers of blood transfusions. N Engl J Med  1978; 299: 799– 803. Google Scholar CrossRef Search ADS PubMed  17 Torrance H D, Brohi K, Pearse R M et al.   Association between gene expression biomarkers of immunosuppression and blood transfusion in severely injured polytrauma patients. Ann Surg  2015; 261: 751– 9. Google Scholar CrossRef Search ADS PubMed  18 Escobar G A, Cheng A M, Moore E E et al.   Stored packed red blood cell transfusion up-regulates inflammatory gene expression in circulating leukocytes. Ann Surg  2007; 246: 129– 34. Google Scholar CrossRef Search ADS PubMed  19 Liu Y, Zeng G. Cancer and innate immune system interactions: translational potentials for cancer immunotherapy. J Immunother  2012; 35: 299– 308. Google Scholar CrossRef Search ADS PubMed  20 Atzil S, Arad M, Glasner A et al.   Blood transfusion promotes cancer progression: a critical role for aged erythrocytes. Anesthesiology  2008; 109: 989– 97. Google Scholar CrossRef Search ADS PubMed  © The Authors 2017. Published by Oxford University Press on behalf of International Society for Diseases of the Esophagus. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

Journal

Diseases of the EsophagusOxford University Press

Published: Feb 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off