Blood transfusion does not affect survival of gastric cancer patients

Blood transfusion does not affect survival of gastric cancer patients

j o u r n a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 5 ) 1 e7 Available online at ScienceDirect journal homepage: w...

742KB Sizes 3 Downloads 100 Views

j o u r n a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 5 ) 1 e7

Available online at

ScienceDirect journal homepage:

Blood transfusion does not affect survival of gastric cancer patients Jingli Cui, PhD, Jingyu Deng, MD, Xuewei Ding, MD, Li Zhang, PhD, Rupeng Zhang, MD, Weipeng Wu, PhD, Xishan Hao, MD, and Han Liang, MD* Department of Gastroenterology, Cancer Hospital of Tianjin Medical University, Key Laboratory of Cancer Prevention and Therapy, and National Clinical Research Center of Cancer, Tianjin, China

article info


Article history:

Background: To initially assess the impact of perioperative blood transfusions (PBTs) on

Received 6 December 2014

overall survival of patients underwent curative resection of ⅠeⅢ TNM stage gastric cancer

Received in revised form

(GC) using the propensity scoring method.

2 July 2015

Methods: The medical records of 1150 GC patients who underwent curative resection in the

Accepted 9 July 2015

Tianjin Cancer Hospital between 2003 and 2008 were retrospectively analyzed. Both

Available online xxx

transfusion and nontransfusion patients were assessed the prognostic differences after surgery using the propensity score analysis.


Results: A total of 299 GC patients (26.0%) were administrated the PBT. With the unadjusted

Gastric cancer

analysis, patients with PBT presented older age, more operative blood loss, lower hemoglobin,

Blood transfusion

lower albumin level, and higher risk of the advanced disease. The 5-y survival rate for patients

Propensity score analysis

with PBT was 31.0%, which was significantly lower than that (47.9%) of patients without PBT


(P < 0.05). However, we demonstrated that there was not any statistical 5-y survival rate difference of between patients with PBT and patients without PBT with the propensity score analysis (31.0% versus 31.3%, P > 0.05). In addition, we also found that PBT was not significantly associated with the increasing risk of mortality (hazard ratio, 1.054; P ¼ 0.628). Conclusions: PBT could not give rise to the worse prognoses of GC patients. ª 2015 Elsevier Inc. All rights reserved.



The perioperative blood transfusion (PBT) has been an effective treatment method for patients with lower hemoglobin or a large amount of blood loss during operation [1,2]. It has been reported that PBT was associated with poorer prognosis for patients with many solid tumor malignancies, including hepatocellular carcinoma, bladder cancer, colorectal cancer, and esophageal cancer [3e6]. Since Burrows and Tartter first reported an unfavorable prognosis in colon cancer patients who perioperatively received blood transfusions, several studies

demonstrated an association between blood transfusion and worse survival. However, other researches held the opposite viewpoint [710]. The large recent meta-analysis of 12,127 patients also failed to answer the question whether there is any observed association between blood transfusion and decreased overall survival (OS). Until now, the impact of PBT on OS for gastric cancer (GC) patients underwent curative resection is still controversial. We therefore used data of 1150 patients with stage IaeIIIc GC in a larger volume center to explore the impact of PBT on long-term outcome. Patients were categorized into two groups

* Corresponding author. Department of Gastric Cancer, Tianjin Medical University Cancer Institute and Hospital, National Clinical Research Center of Cancer, Key Laboratory of Cancer Prevention and Therapy, Tianjin 300060, China. Tel./fax: þ86 22 23340123. E-mail address: [email protected] (H. Liang). 0022-4804/$ e see front matter ª 2015 Elsevier Inc. All rights reserved.


j o u r n a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 5 ) 1 e7


Materials and methods

laboratory tests. The PBT was defined as transfusion of erythrocytes during perioperative hospitalization. Notably, administration of PBT was based on the discretion of the treating physician. No institutional standardized intraoperative or postoperative thresholds were used for transfusion. In general, patients whose perioperative hemoglobin was <70 g/L or who lost a lot of blood during surgery were routinely given a red blood cell transfusion.




based on transfusion status, the transfused group, and nontransfused group. A one-to-one propensity score matching method was used to overcome bias due to the different distributions of covariates for the two groups. Clinicopathologic featureerelated prognostic factors were analyzed.

Surgical management

A total of 1542 patients with histologically confirmed gastric adenocarcinoma who underwent curative surgery in Tianjin Medical University Cancer Institute and Hospital between January 2003 and December 2008 were eligible for this study. Eligibility criteria included (1) adenocarcinoma of the stomach, (2) patients with stage IeIII disease who underwent gastrectomy with curative intent, (3) no history of gastrectomy with curative intent, (4) no history of gastrectomy or other malignancy, and (5) no history of neoadjuvant chemotherapy. In hospital mortality, 23 patients were excluded from the analysis. Also, 219 patients with distant metastases at the time of operation and 41 patients undergoing R1 procedures were excluded. Finally, 67 patients were excluded due to lost to follow-up, 25 patients due to missing hemoglobin levels, and 12 patients due to missing data regarding blood transfusions (Fig. 1). Ultimately, a total of 1150 GC patients were included in this study. Patients were divided in two groups: patients who did and those who did not receive PBTs.

All patients underwent curative gastrectomy plus lymphadenectomy. Curative resection was defined as complete absence of grossly visible tumor tissue and pathologically negative resection margins. Primary tumors were resected en bloc with lymphadenectomy (D1 or D2) according to the guidelines of the Japanese Gastric Cancer Association. The surgical procedures were based mainly on the GC treatment guidelines in Japan.



Data collection and definitions

Clinicopathologic data of all patients were included in this study. They were as follows: age at surgery, gender, body mass index (BMI), preoperative hemoglobin level, PBT, operative blood loss, TNM classification, histologic results, lymph-node metastasis, adjuvant chemotherapy, and blood biochemical

Fig. 1 e The criteria for inclusion and exclusion of all patients.



After undergoing curative surgery, all patients were followed up every 3 or 6 mo for 2 y and annually thereafter until death. The median follow-up for the entire cohort was 40 mo. Followup of all patients included in this study was completed in December 2013. Ultrasonography, computed tomography scans, chest x-rays, and endoscopy were performed at every visit.

Statistical analysis

Statistical analyses were performed using the SPSS statistical software version 19 (SPSS, Inc, Chicago, IL). A two-sided P < 0.05 was considered statistically significant. Continuous data are expressed as means  standard deviation. For comparing proportions, chi-square statistics, and for comparing continuous variables, t-tests were used. The KaplaneMeier method was used for analysis of OS, whereas comparisons between different groups were performed using the log-rank test. To overcome bias due to the different distribution of covariates among patients from the two groups (patients with PBT and patients without PBT), a propensity score analysis as a superior statistical method of adjusting for potential baseline confounding variables was performed [7,11]. A propensity score analysis as a superior and more refined statistical method of adjusting for potential baseline confounding variables was performed. The “MatchIt” and the “optmatch” R packages were used to perform a bipartite weighting propensity score analysis. The distance measure was estimated by logistic regression using the risk set described previously to predict blood transfusions. Patients receiving blood transfusions without a counterpart regarding the distance measure among the patients not receiving blood transfusions and vice versa were excluded from the analysis. Thereafter, the distance measure was reestimated. Otherwise, the default settings were left unchanged. The baseline risk profiles of the matched patients were compared to assure that no major differences in baseline patients’ characteristics persisted. The prognostic value of blood transfusion on OS was finally

j o u r n a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 5 ) 1 e7


assessed in a mixed-effects Cox regression model by applying the subclasses and weights obtained by the propensity score analysis using the R package “coxme.” Then, the prognostic value of blood transfusions for overall and disease-free survival was assessed using a Cox regression analysis of the matched data with adjustments for the subclasses and weights that were obtained in the propensity score matching procedure. Variables entered in the propensity model were age at surgery, sex, BMI, white blood cells, platelet, albumin, total protein, creatinine, urea, tumor diameter, histologic, type of gastrectomy, extent of lymphadenectomy, chemotherapy, depth of invasion (T stage), lymph-node metastasis (N stage), TNM classification, and operative factors (excluding operative blood loss and hemoglobin level). The propensity score analysis was used to obtain a one-to-one match using the nearest-neighbor matching method [12,13] (shown in Fig. 2 and Fig. 3).




Patient characteristics and blood transfusions

Of 1150 patients, 229 (26.0%) were administrated with PBT and 851 patients were administrated without PBT. Clinical pathologic demographics for these two groups of patients are shown in Table 1. Patients with PBT were significantly older (62.0 versus 60.5 y, P ¼ 0.042), had lower BMI (22.8 versus 23.4, P ¼ 0.026), lower hemoglobin levels (107.5 versus 130.6 g/dL, P < 0.001), lower serum albumin levels (38.7 versus 41.8, P < 0.002), higher platelet level (262.3 versus 243.3 g/dL, P < 0.002), greater maximum tumor diameter (6.1 versus 5.0 cm, P < 0.001), greater intraoperative blood loss (232.0 versus 164.5, P < 0.001), and with more advanced TNM stage (P < 0.001) than patients without PBT (Table 1).

Fig. 2 e The distribution of the propensity scores. Each circle represents one patient. The distributions of the propensity scores for all the patients who could be matched are shown in the vertical center of the figure. The sizes of the circles for the matched patients who did not receive blood transfusions are proportional to the weights that were obtained in the propensity score matching procedure.

Fig. 3 e Standardized difference before and after matching.

Median postoperative follow-up for all patients in this study was 41.0 mo, whereas the patients with PBT and patients without PBT were 26.0 and 28.0 mo, respectively. In the entire population, survival rate of patients with PBT was significantly lower than that of patients without PBT (c2 ¼ 33.420, P < 0.000). In the patients with PBT, the 3- and 5-y survival rates (5-YSRs) were 41.8% and 31.0%, respectively. However, in the patients without PBT, 5-YSRs were 56.4% and 47.9%, respectively (Fig. 4). As shown in Table 2, multivariate analysis of the variables that were determined to be significant for OS by univariate analysis in the whole study series revealed that PBT (hazard ratio [HR], 1.246; 95% confidence interval [CI], 1.025e1.515; P ¼ 0.0.027), age (HR, 1.012; 95% CI, 1.004e1.020, P ¼ 0.007), tumor size (HR, 1.059; 95% CI, 1.023e1.096; P ¼ 0.001), serum albumin (HR, 0.980; 95% CI, 0.963e0.998; P ¼ 0.026), T-stage (HR, 1.361; 95% CI, 1.170e1.584; P < 0.001), N-stage (HR, 1.550; 95% CI, 1.434e1.673; P < 0.001), extent of lymphadenectomy (HR, 1.364; 95% CI, 1.144e1.625, P ¼ 0.002), chemotherapy (HR, 0.769; 95% CI, 0.645e0.916; P ¼ 0.003), and operative blood loss (HR, 3.206; 95% CI, 1.414e7.284; P ¼ 0.005) are independent prognostic factors for OS.


Results after propensity score match

In this study, we selected 229 patients with PBT as the control group by one-to-one matching using propensity scores. Characteristics after the propensity score analysis are shown in Table 1. A total of 229 patients with PBT were matched with 229 patients without PBT after covariate


j o u r n a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 5 ) 1 e7

Table 1 e Clinicopathologic features of GC patients who underwent gastrectomy with curative intent grouped by blood transfusion. Variables

Gender Male/female Age (y) BMI (kg/m2) Hemoglobin (g/L) WBC PLT Albumin (g/L) Total protein (g/L) Creatinine (g/L) Urea (g/L) Tumor diameter (cm) Operative blood loss (mL) Histologic grading Well or moderately/poorly Type of gastrectomy Total/subtotal Extent of lymphadenectomy D1 or D1þ / D2 or D2þ Chemotherapy No/yes Depth of invasion, n (%) pT1 pT2 pT3 pT4 Lymph-node metastasis, n (%) pN0 pN1 pN2 pN3 pTNM stage, n (%) I II III

Whole study series

Propensity-score-matched pairs

Nontransfused, n ¼ 851

Transfused, n ¼ 299

604/247 60.5 (26e93) 23.4 (10.4e48.8) 130.6 (51e188) 6.6 (2.6e61.9) 243.3 (2e621) 41.8 (4.0e78.8) 70.4 (4.8e96.2) 69.3 (1.0e173.0) 6.3 (0.1e331.0) 5.0 (0.2e35.0) 164.5 (10e800)

213/86 62.0 (23e89) 22.8 (9.4e37.6) 107.5 (8e169) 6.7 (1.8e103.0) 262.3 (2e678) 38.7 (26.5e52.4) 69.1 (25.3e559.4) 68.3 (8.0e208.0) 5.9 (1.7e66.0) 6.1 (0.3e25.0) 232.0 (20e3000)















0.063 <0.001



Nontransfused, n ¼ 299

0.931 0.042 0.026 <0.001 0.647 0.002 <0.001 0.270 0.420 0.582 <0.001 <0.001

P 0.857

212/87 62.7 (31e84) 22.6 (13.2e39.7) 125.1 (51e170) 6.6 (2.6e17.79) 268.1 (3e621) 40.0 (4e78.8) 67.5 (5.5e96.2) 69.1 (5.8e173.0) 5.8 (1.1e65.0) 6.1 (0.3e35.0) 170.6 (10e800)

0.427 0.398 <0.001 0.773 0.405 0.125 0.410 0.669 0.818 0.898 <0.001 0.212 0.724 0.999 0.412

29 105 90 627

(3.4) (12.3) (10.6) (73.7)

4 16 23 256

(1.3) (5.4) (7.7) (85.6)

345 157 181 168

(40.5) (18.5) (21.3) (19.7)

90 58 72 79

(30.1) (19.4) (24.1) (26.4)

0.848 2 18 23 256

(0.7) (6.0) (7.7) (85.6)

72 65 84 78

(24.1) (21.7) (28.1) (26.1)



<0.001 101 (11.9) 278 (32.7) 472 (55.4)

13 (4.3) 85 (28.4) 201 (67.3)

0.853 14 (4.7) 79 (26.4) 206 (68.9)

PLT ¼ platelet; WBC ¼ white blood cell. Data are reported for the whole study series and for one-to-one propensity score matched pairs. P values less than 0.05 were considered statistically significant.

adjustment. The study group of 458 patients was well matched: all covariates that significantly affected OS in the entire study group were equally distributed over the two matched groups except the only two variables, hemoglobin level and operative blood loss, which were closely related with PBT. So the clinical variables and tumor characteristics were similar for transfused and nontransfused patients. In the matched study group, the median follow-up was 27 and 26 mo of nontransfused and transfused groups. The OS rate for patients with PBT was similar to that of patients without PBT (c2 ¼ 0.231, P ¼ 0.631). The 3- and 5-YSRs in the patients without PBT were 42.9% and 31.3%, and those in the patients with PBT were 42.0% and 31.0% (Fig. 5). As shown in Table 2, multivariate analysis of the variables that were determined to be significant for OS by univariate analysis in the propensity scoreematched pairs revealed that N-stage and operative blood loss were independent prognostic factors for OS rather than PBT.



In view of D2 lymph-node dissection leading more blood loss [14], PBT is usually required for GC patients. However, the impact of PBT on OS of GC patients remains controversial. Several authors reported that PBT promoted the recurrence of GC and decreased OS after surgery. Kaneda et al. [15] proposed that PBT could have an adverse effect on survival of GC patients in 1987. Ojima et al. [16] reported that the 5-y diseasefree survival rate was significantly worse in the patients with PBT and identified that PBT was an independent prognostic factor for long-term survival patients. On the other hand, several investigators showed that there is no significant correlation between PBT and prognosis of GC patients. Kampscho¨er et al. [17] showed no 5-YSRs difference between patients with PBT and patients without PBT. Moriguchi et al. [18] also failed to reveal the relationship between


j o u r n a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 5 ) 1 e7

Fig. 4 e The 5-YSR was 47.9% for whole study series without non-transfused and 31.0% for those transfused (P < 0.001).

PBT and OS of GC patients. Rausei et al. [19] showed that PBT seemed a confounding factor more than a prognostic indicator because it was obviously affected by other clinicopathologic variables. They further described that PBT were closely associated with other prognostic covariates, and there was no prognostic significance of blood transfusions on survival time. It is a fact that the administration of blood products can cause profound negative effects on the human immune system, a condition termed transfusion-related immune modulation, although its effects on long-term survival in patients with GC after curative surgery remain controversial. The mechanisms for transfusion-related immune modulation include suppression of cytotoxic cell and monocyte activity, release of immunosuppressive prostaglandins, inhibition of interleukin-2 production, and increase in suppressor T-cell activity [20,21]. Several studies demonstrated that the potentially adverse influence of PBT on the survival of GC patients may be associated to other prognostic features rather than the immunologic sequelae of the transfusion itself [17]. To our knowledge, this is the first study to investigate the influence of PBT on GC patients using propensity score analysis. The present study demonstrated that when other prognosis variables were appropriately adjusted, PBT had no impact on the OS of GC patients after surgery. The result of this study contradicted previous studies that reported that

Table 2 e Multivariate survival analysis of GC patients in the whole study series and propensity scoreematched pairs. Prognostic factors



Whole study series

Propensity scoreematched pairs

HR (95% CI) Blood transfusion Yes/no Sex Male/female Age (y) 60/>60 BMI (kg/m2) 25/>25 Tumor size, cm 5/>5 Hemoglobin (g/dL) Anemia/no anemia Albumin 35/>35 Histologic grading Well/poorly T stage T1/T2/T3/T4 N stage N0/N1/N3 Extent of lymphadenectomy D2/D1 Type of gastrectomy Subtotal/total Chemotherapy Yes/no Operative blood loss (g) 300/>300


HR (95% CI)


1.246 (1.025e1.515)


1.054 (0.852e1.305)


0.828 (0.679e1.009)


0.825 (0.652e1.043)


1.012 (1.004e1.020)


1.159 (0.926e1.451)


1.004 (0.981e1.028)


1.038 (0.818e1.317)


1.059 (1.023e1.096)


1.140 (0.917e1.416)


1.003 (1.000e1.007)


0.991 (0.796e1.233)


0.980 (0.963e0.998)


0.738 (0.589e0.926)


1246 (1.025e1.515)


1.091 (0.954e1.247)


1.361 (1.170e1.584)


1.351 (1.006e1.815)


1.550 (1.434e1.673)


1.508 (1.255e1.812)


1.364 (1.144e1.625)


0.801 (0.619e1.038)


1.241 (0.993e1.485)


1.191 (0.903e1.570)


0.769 (0.645e0.916)


0.966 (0.909e1.026)


3.206 (1.411e7.284)


3.297 (1.170e9.289)



j o u r n a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 5 ) 1 e7

Disclosure The authors claim that none of the material in the article has been published or is under consideration for publication elsewhere. The authors state no conflict of interest and funding sources exit in the submission of this article.


Fig. 5 e The 5-YSR was 31.3% for the matched study series without non-transfused and 31.0% for those transfused (P [ 0.631).

PBT was a risk factor for prognosis in GC patients. In our study, there was not any significant difference between patients with PBT and patients without PBT with respect to the clinical background and the operative data (excluding hemoglobin level and operative blood loss). Therefore, we concluded that the difference of OS between the two groups of GC patients before matching, due to the perioperative clinical characteristics and operative course. In gastrectomy, only the aforementioned study described a significant difference in the survival rates according to the amount of PBT [22]. However, our results demonstrated that PBT could not affect the prognosis of GC patients regardless of the amount of transfused blood.

Acknowledgment This work was supported by National Basic Research Program of China (973 Program) (no. 2010CB529301), the Key Program of Tianjin Municipal Science and Technology Commission (13ZCZCSY20300), and the National Science and Technology Major Projects for “Major New Drugs Innovation and Development” (no. 2013ZX09303001). Authors’ contributions: J.C. and J.D. contributed to the conception. J.C. contributed to the designs and draft of the work and revised it critically for important intellectual content. X.D., L.Z., R.Z., W.W., J.D., and H.L. did the acquisition, analysis, or interpretation of data of the work. X.H. and H.L. approved the version to be published. X.H. agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

[1] Zilberstein B, Jacob CE, Cecconello I. Gastric cancer trends in epidemiology. Arq Gastroenterol 2012;49:177. [2] Sano T, Aiko T. New Japanese classifications and treatment guidelines for gastric cancer: revision concepts and major revised points. Gastric Cancer 2011;14:97. [3] Katz SC, Shia J, Liau KH, et al. Operative blood loss independently predicts recurrence and survival after resection of hepatocellular carcinoma. Ann Surg 2009; 249:617. [4] Morgan TM, Barocas DA, Chang SS, et al. The relationship between perioperative blood transfusion and overall mortality in patients undergoing radical cystectomy for bladder cancer. Urol Oncol 2013;31:871. [5] Harlaar JJ, Gosselink MP, Hop WC, Lange JF, Busch OR, Jeekel H. Blood transfusions and prognosis in colorectal cancer: long-term results of a randomized controlled trial. Ann Surg 2012;256:681. discussion 6e7. [6] Komatsu Y, Orita H, Sakurada M, Maekawa H, Hoppo T, Sato K. Intraoperative blood transfusion contributes to decreased long-term survival of patients with esophageal cancer. World J Surg 2012;36:844. [7] Tarantino I, Ukegjini K, Warschkow R, et al. Blood transfusion does not adversely affect survival after elective colon cancer resection: a propensity score analysis. Langenbecks Arch Surg 2013;398:841. [8] Kuroda S, Tashiro H, Kobayashi T, Oshita A, Amano H, Ohdan H. No impact of perioperative blood transfusion on recurrence of hepatocellular carcinoma after hepatectomy. World J Surg 2012;36:651. [9] Yeoh TY, Scavonetto F, Weingarten TN, et al. Perioperative allogeneic nonleukoreduced blood transfusion and prostate cancer outcomes after radical prostatectomy. Transfusion 2014;54:2175. [10] Kim SH, Lee SI, Noh SM. Prognostic significance of preoperative blood transfusion in stomach cancer. J gastric Cancer 2010;10:196. [11] Zinsmeister AR, Connor JT. Ten common statistical errors and how to avoid them. Am J Gastroenterol 2008;103:262. [12] Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997;127(8 Pt 2):757. [13] D’Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a nonrandomized control group. Stat Med 1998;17:2265. [14] Wagman LD. “Necessary” transfusions and prognosis in gastric cancer. Ann Surg Oncol 2002;9:1. [15] Kaneda M, Horimi T, Ninomiya M, et al. Adverse affect of blood transfusions on survival of patients with gastric cancer. Transfusion 1987;27:375. [16] Ojima T, Iwahashi M, Nakamori M, et al. Association of allogeneic blood transfusions and long-term survival of patients with gastric cancer after curative gastrectomy. J Gastrointest Surg 2009;13:1821. [17] Kampschoer GH, Maruyama K, Sasako M, Kinoshita T, van de Velde CJ. The effects of blood transfusion on the prognosis of patients with gastric cancer. World J Surg 1989;13:637.

j o u r n a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 5 ) 1 e7

[18] Moriguchi S, Maehara Y, Akazawa K, Sugimachi K, Nose Y. Lack of relationship between perioperative blood transfusion and survival time after curative resection for gastric cancer. Cancer 1990;66:2331. [19] Rausei S, Ruspi L, Galli F, et al. Peri-operative blood transfusion in gastric cancer surgery: prognostic or confounding factor? Int J Surg 2013;11:S100.


[20] van Twuyver E, Mooijaart RJ, ten Berge IJ, et al. Pretransplantation blood transfusion revisited. N Engl J Med 1991;325:1210. [21] Blajchman MA. Immunomodulation and blood transfusion. Am J Ther 2002;9:389. [22] Hyung WJ, Noh SH, Shin DW, et al. Adverse effects of perioperative transfusion on patients with stage III and IV gastric cancer. Ann Surg Oncol 2002;9:5.