- Email: [email protected]
Perioperative blood transfusion adversely affects prognosis of patients with stage I non-small-cell lung cancer
Perioperative Blood Transfusion Adversely Affects Prognosis of Patients with Stage I Non-Small-Cell Lung Cancer Alex G. Little, MD, Las Vegas,Nevada,Reno,Nevada, Huai-Shen Wu, MD, Mark K. Ferguson, MI), Chicago,IIIinois, Chih-Hsiang Ho, Phl), Las Vegas,Nevada, Victor D. Bowers, MO, Cas Vegas,Nevada,Reno,Nevada, Andrea Segalin, Mo, Victoria M. Staszek, RN, BSN,Chicago,Illinois
It has been speculated that blood transfusion might adversely affect prognosis in cancer patients by immunosuppression. To avoid the confounding affect of advanced disease, we tested this hypothesis in 117 patients with stage I non-small-cell lung cancer. Mean and median follow-up were 49.7 months and 47 months, respectively. Patients who died during the postoperative period were not included. Perioperative transfusion was defined as administration of whole blood or packed cells within 30 days of operation. The overall cumulative 5-year disease-free survival rate was 67%. In patients with transfusion, it was 53% and in patients without transfusion it was 8 I % (p = 0 . 0 0 5 5 ) . A multivariate analysis was performed that included patient age, race, sex, cell type, extent of operation (pneumonectomy versus lobectomy/segmenteetomy), operative blood loss, admission hematocrit, discharge hematocrit, and the presence or absence of perioperative transfusion. The only variable that significantly correlated with 5-year disease-free survival was the presence or absence of perioperative transfusion (p = 0 . 0 2 7 8 ) , and this effect was not related to the n u m b e r of transfusions. Retrospective analysis of long-term results of patients surviving curative operation for stage I lung cancer shows that any perioperative transfusion significantly worsens the patient's prognosis and suggests very strongly that this association is due to an adverse effect of the transfusion rather than the transfusion serving as a marker for another risk factor.
From the Department of Surgery (AGL, VDB), University of Nevada School of Medicine, Las Vegas, Nevada, and Reno, Nevada; the Department of Surgery (HSW, MKF, AS, VMS), University of Chicago, Pritzker School of Medicine, Chicago, Illinois; and the Department of Mathematical Sciences (CHH), University of Nevada, Las Vegas, Nevada. Requests for reprints should be addressed to Alex G. Little, MD, Department of Surgery, University of Nevada School of Medicine, 2040 West Charleston Boulevard, Suite 601, Las Vegas, Nevada 89102. Presented at the 42nd Annual Meeting of the Southwestern Surgical Congress, La Quinta, California, April 22-25, 1990.
630
THE AMERICAN JOURNAL OF SURGERY
ecognition that blood transfusion enhances graft survival in patients undergoing kidney transplantaR tion focused attention on the immunosuppressive effects of blood components. Analysis of patients undergoing operation for cancer has revealed varying results for transfusion effects in patients with tumors of different organ systems. However, substantial data exist to suggest very strongly that blood transfusion and its subsequent immunosuppression has a significant, deleterious effect on patients with early stage cancer undergoing surgical resection [1]. In particular, several reports of patients undergoing surgical resection for lung cancer exist [2-5]. Although they contain some conflicting results and conclusions, most have suggested a deleterious transfusion effect in this patient population. Several of these reports are difficult to interpret because of the inclusion of patients with a relatively advanced disease stage. Because the prognosis of these patients is worse than for patients with earlierstage disease, it is more difficult to show an effect of blood transfusion. To further analyze this important issue in thoracic surgical oncology, we reviewed a series of 117 patients with stage I (T1,N0,M0 or T2,N0,M0) nonsmall-cell lung cancer to provide additional data concerning the effect of blood transfusion in these patients. PATIENTS AND M E T H O D S The study includes 117 patients with stage I nonsmall-cell lung cancer who received operative therapy and in whom complete follow-up was available either until death or until completion of the study. These patients were operated on either at the University of Chicago Hospitals, Chicago, Illinois, or at University Medical Center, Las Vegas, Nevada. There were 60 male and 57 female patients, with 54 white, 60 black, and 3 Asian patients. Mean and median follow-up for all patients was 49.7 months and 47 months, respectively. We did not include patients who died during the perioperative period to allow examination of the effect of blood transfusion on long-term survival. Patients were divided into two groups depending upon their transfusion history. Patients who received administration of either whole blood or packed red blood cells within 30 days before or after operation constituted the transfusion group. The remaining patients formed the no transfusion group. Statistical comparisons of the two groups were performed with chi-square analysis or analysis of variance. Survival analysis was performed using the Statistical Analysis System and the Statistical Package for the Social Sciences.
VOLUME 160
DECEMBER 1990
EFFECT OF PERIOPERATIVE BLOOD TRANSFUSION ON CANCER
TABLE I
O1
PerloperaUve Transfusion In Stage I Non-Small-Cell Lung Cancer (mean 4- SD) Transfusion (n = 58)
No Transfusion (n = 59)
,9~ .8| .7~
p Value
Patient 61.3 4- 8.8 60.1 4- 9.6 0.4719 Age (yr) M 32, F 26 M 28, F 31 NS Sex White 31, Black 31 White 29, Black 29 NS Race T1 31, T2 27 T1 34, T2 25 NS Primary tumor Cell type 26 (45%) 37 (63%) 0.0793 Adenocarcinoma 25 (43%) 16 (27%) 0.1056 Squamous cell carcinoma 7 (12%) 6 (10%) 0,7438 Large cell carcinoma Operation 10 (17%) 8 (14%) 0.7675 Pneumonectomy Other 48 (83%) 51 (86%) 0.5810 578.0 4- 488.6 320.4 4- 198.8 0.0003* Blood loss (mL) Hematoorit 37.7 4- 5.2 39.8 4- 4.5 0.0248" Admission 34.5 4- 4.6 34.8 4- 4,9 0.7897 Discharge * Difference significant.
i
.64 .5| .4 4
NO T x T.
0
20
40
60
so
100 12o 140 16o 18o 200
Months
Figure 1. This figure contrasts the disease-free survival curve for the 58 patients who received transfusion (TX) to the survival curve of 59 patients who did not receive transfusions. The difference in 5-year disease-free survival, as described in the text, is significant with p = 0.0055.
TABLE
II
5 - Y e a r Disease-Free Survival (mean -I- 1 SE)
RESULTS There were 117 patients with histologically confirmed non-small-cell lung cancer. These patients were operated on between January 1, 1977, and December 31, 1986. Sixty-three patients had adenocarcinoma, 41 patients had squamous cell carcinoma, and 13 patients had large cell carcinoma. Eighteen patients had pneumonectomy as their operation and 99 patients had either lobectomy, segmentectomy, or wedge resection as their surgical treatment. Preoperative staging varied but, in addition to history and physical examination, included chest radiographs in all patients, computed tomographic scans of the chest and upper abdomen in all patients after 1981, mediastinoscopy in 64 patients, and a variety of radionuclide scans in many patients. Final pathologic staging documented stage I disease in all patients, most of whom had systematic removal of multiple mediastinal lymph node basins during their operation. A total of 58 patients received transfusion, constituting the perioperative transfusion group, and 59 patients received no transfusion. Table I compares the transfused with the non-transfused groups. The only significant differences are in the average operative blood loss and the admission hematocrit. Computation of survival functions was based on the actuarial method of Berkson and Gage [6] and was performed using the Statistical Package for the Social Sciences. The overall cumulative 5-year disease-free survival rate was 67%. A statistic was calculated from the survival scores using the algorithm of Lee and Desu [7]. In patients with transfusion, 5-year disease-free survival was 53% and in patients without transfusion it was 81% (p = 0.0055) (Figure 1). Table II shows the effect on survival of multiple variables when considered independently.
Cumulative Survival (%) All patients (n = 117) Transfusion (n = 58) No transfusion (n = 59) Male (n = 60) Female (n = 57) White (n --- 54) Black (n --- 60) Adenocarcinoma (n = 63) Squamous (n = 41) Pneumonectomy (n = 18) Other operations (n = 99)
67 53 81 59 74 63 70 76 81 67 67
4- 0.051 4- 0.073 4- 0.066 4- 0.080 4- 0.065 4- 0.073 4- 0.072 4- 0.068 4- 0.80 4- 0.111 4- 0.057
p Value
-0.0055" 0.1765 0.3323 0.0143" 0.1428
* Difference significant.
A parametric accelerated failure time or regression model was fitted to the survival time using the Statistical Analysis System. The covariates are patient age, patient race, patient sex, cell type, extent of operation (pneumonectomy versus lobectomy/segmentectomy), operative blood loss, admission hematocrit, discharge hematocrit, and the presence or absence of perioperative transfusion. The model indicates that the only covariate that significantly correlates with 5-year disease-free survival is the presence or absence of perioperative transfusion (p = 0.0278). Table III shows that transfusion is associated with a worse prognosis for all patient subgroups. Interestingly, this correlation is greater in women than in men, in black than in white patients, and in patients with squamous cell carcinoma than in patients with adenocarcinoma. In addition, this effect is not related to the quantity of blood transfused since survival in patients receiving one or two units of blood (50%) is similar to that in patients receiving three or more units (59%) (p = 0.3043).
THE AMERICAN JOURNAL OFSURGERY
V O L U M E 160
D E C E M B E R 1990
631
LITTLE ET AL
TABLE III 5-Year Disease-Free Survival In Patient Subgroups Transfusion (%) All patients Males Females White Black Adenocarcinoma Squamous cell carcinoma
Pneumonectomy Other operations Transfusion 1-2 units 3 or more
53 47 60 48 59 64 49 50 55
44444444. 4-
0.073 0.107 0.100 0.108 0.100 0.107 0.105 0.110 0.106
50 4- 0.100 59 4- 0.104
No Transfusion (%) 81 71 90 79 83 85 80 88 80
44444444. 4-
0.066 0.116 0.055 0.077 0.099 0.080 0.103 0.105 0.090
p Value 0.0055 0.0851 0.0265 0.0699 0.0344 0.1114 0.0617 0.1145 0.0348 0.3043
COMMENTS Transfusion of blood components clearly has an immunosuppressive effect. Exposing a patient to blood product antigens stimulates the immune system to some extent, and the early immunology literature documented and emphasized this effect [8]. However, experience with and analysis of patients undergoing renal transplantation have unequivocally shown that perioperative transfusion improves graft survival through a nonspecific immunosuppressive action [9,10]. This immunosuppressive effect is incompletely explicated, but the data show that transfusion increases the number and/or activity of suppressor T lymphocytes, decreases the number of natural killer ceils, induces the formation of anti-idiotypic antibodies, and impairs lymphocyte blastogenesis [11-14] With this documented suppressive effect of blood components on the immune system, it is not surprising that there is the potential for a deleterious effect in patients undergoing operative treatment for cancer. This effect seems to vary between cancers arising in different organs. In lung cancers, several reports exist with varying results [2-5]. Unfortunately, interpretation of these studies is complicated by their inclusion of patients with relatively advanced disease, specifically patients with stage II and stage III lung cancers. In these patients, whose overall prognosis is considerably poorer than patients with stage I disease, it is harder to identify a real but modest effect on prognosis due to blood transfusions. Three of the four series, however, did identify a statistically significant deleterious effect on prognosis in transfused patients with stage I and/or II non-small-cell lung cancer undergoing resectional therapy, For this reason, we focused completely on patients with early stage non-small-cell lung cancer. Although a retrospective analysis like ours does not prove a clinically significant effect, the definite statistical correlation does indicate this conclusion very strongly, particularly when the two patient groups are so similar. There is a statistically significant difference in the admission hematocrit level but it is unlikely that this difference, 37.7 4- 5.2 versus 39.8 4- 4.5, is clinically significant. Independent comparison of variables suggests that both 632
transfusion and cell type correlate with survival. However, only the presence or absence of perioperative transfusion is a significant covariate in the regression model. The suggestions that this effect is more pronounced in women than in men, in black than in white patients, and in patients with squamous cell carcinoma than those with adenocarcinoma are interesting but require future corroboration by other investigators. Given the findings of this study, which strengthen similar conclusions from earlier investigations, it is clear that perioperative and intraoperative care of patients undergoing surgical resection of stage I non-small-cell lung cancer should include concerns about the effect of transfusion. A patient should not undergo elective transfusion simply because of a moderately low hemoglobin or hematocrit level. The need for transfusion should be based solely on specific indications such as evidence of inadequate tissue oxygenation. Under most circumstances, the patient should be allowed to replenish red blood cell and hemoglobin levels endogenously, aided by oral iron supplementation if necessary. Reinfusion of autologous blood during surgery is probably inadvisable because of the risk of contamination with cancer cells.
REFERENCES 1. Wu HS, Little AG. Perioperative blood transfusions and cancer recurrence. J Clin Oncol 1988; 6: 1348-54. 2. Tartter PI. Burrows L. Kirschner P. Perioperative blood transfusion adversely affects prognosis after resection of stage I (subset N O ) non-oat cell lung cancer. J Thorac Cardiovasc Snrg 1984; 88: 659-22. 3. Hyman NH, Foster RS, DeMeules JE, Costanza MC. Blood transfusions and survival after lung cancer resection. Am J Surg 1985; 149: 502-7. 4. Keller SM, Groshen S, Martini N, Kaiser LR. Bloodtransfusion and lung cancer recurrence. Cancer 1988; 62: 606-10. 5. Moores DW, Piantadosi S, McKneally MF. Effect of perioperarive blood transfusion on outcome in patients with surgically resected lung cancer. Ann Thorac Surg 1989; 47: 346-51. 6. BerksonT, Gage R. Calculation of survivalrates for cancer. Proc Mayo Clin 1950; 25: 270-6. 7. Lee E, Desu M. A computer program for comparing k samples with right-censored data. Comp Prog Biomed 1972; 2: 315-21. 8. Medawar PE. Immunityto homologousskin II. The relationship between antigens on blood and skin. Br J Exp Pathol 1946; 27: 1521. 9. Opelz G, Sengor DPS, McKey MR, Terasaki PI. Effect of blood transfusions on subsequent kidney transplants. Trans Proc 1973; 5: 253-9. 10. Opelz G, Terasaki PI. Improvement of kidney graft survival with increased numbers of blood transfusions. N Engl J Med 1978; 299: 799-804. 11. Fischer E, Lenhard V, Seifect P, et al. Blood transfusioninduced suppression of cellular immunity in man. Hum Immunol 1980; 1: 187-94. 12. Smith MD, Williams JD, Coles GA, et al. The effect of blood transfusion on T-suppressor cells in renal dialysis patients. Transplant Proc 1981; 13: 181-3. 13. Kaplan J, Sarnaik S, Gitlin J, et al. Diminishedhelper/suppressor lymphocyte ratios and natural killer activity in recipients of repeated blood transfusion. Blood 1984; 64: 308-10. 14. Gasim P, Zoumbos NC, Young NS. Immunologic abnormalities in patients receiving multiple blood transfusions. Ann Intern Med 1984; 100: 173-7.
THE AMERICAN JOURNAL OF SURGERY VOLUME 160 DECEMBER 1990
EFFECT OF PERIOPERATIVE BLOOD TRANSFUSION ON CANCER
DISCUSSION William F. Sasser (St. Louis, MO): This effect is presumed to be due to immunosuppression. Tumor size and lymph node involvement are the most reliable prognostic indices in lung cancer, although blood transfusion should be considered a significant factor for recurrence and survival similar to an elevated carcinoembryonic antigen, the cancer cell type, the lowered serum protein, and the elevated white blood cell count. In the future, molecular studies will solve the theoretical problem. The practical solution now is to restrict and prevent blood transfusions. There were several questions that I would like to ask. In a report of colon resections, patients who received blood transfusions had an increase in postoperative infections; did you observe any increase in sepsis or infection in the patients who received transfusions? The perioperative transfusion effect was 30 days before and 30 days after surgery. If preoperatively your patient was so anemic that blood was required, would you delay surgery 30 days? One requirement for transfusion as stated in the manuscript was proper tissue oxygenation. What
was your criterion for this? Is there a role for autologous blood collection 2 weeks before the operation in patients with stage I lung cancer? Are there any data to indicate that the patients' stored blood is altered such that it has an immunosuppressive influence when reinfused? Alex G. Little (closing): We did not look at the rate or incidence of sepsis in these patients. Is 30 days reasonable? I came up with 30 days simply because that is what other series have used. I doubt I would want to delay surgery 30 days after a transfusion in a patient with cancer just because we know that is a dangerous time period. Investigators who have looked at the duration of the immunosuppressive effect have found that it peaks at approximately 2 weeks. Again, I believe we should only perform transfusions when there are concerns about tissue oxygenation and perfusion. And finally, I think reinfusing shed blood has theoretic possibilities; however, even in patients with limited-stage cancer, it has been shown that malignant cells can enter the pleural space during surgery.
THE AMERICAN JOURNAL OF SURGERY VOLUME160 DECEMBER 1990 633
It has been speculated that blood transfusion might adversely affect prognosis in cancer patients by immunosuppression. To avoid the confounding affect of advanced disease, we tested this hypothesis in 117 patients with stage I non-small-cell lung cancer. Mean and median follow-up were 49.7 months and 47 months, respectively. Patients who died during the postoperative period were not included. Perioperative transfusion was defined as administration of whole blood or packed cells within 30 days of operation. The overall cumulative 5-year disease-free survival rate was 67%. In patients with transfusion, it was 53% and in patients without transfusion it was 8 I % (p = 0 . 0 0 5 5 ) . A multivariate analysis was performed that included patient age, race, sex, cell type, extent of operation (pneumonectomy versus lobectomy/segmenteetomy), operative blood loss, admission hematocrit, discharge hematocrit, and the presence or absence of perioperative transfusion. The only variable that significantly correlated with 5-year disease-free survival was the presence or absence of perioperative transfusion (p = 0 . 0 2 7 8 ) , and this effect was not related to the n u m b e r of transfusions. Retrospective analysis of long-term results of patients surviving curative operation for stage I lung cancer shows that any perioperative transfusion significantly worsens the patient's prognosis and suggests very strongly that this association is due to an adverse effect of the transfusion rather than the transfusion serving as a marker for another risk factor.
From the Department of Surgery (AGL, VDB), University of Nevada School of Medicine, Las Vegas, Nevada, and Reno, Nevada; the Department of Surgery (HSW, MKF, AS, VMS), University of Chicago, Pritzker School of Medicine, Chicago, Illinois; and the Department of Mathematical Sciences (CHH), University of Nevada, Las Vegas, Nevada. Requests for reprints should be addressed to Alex G. Little, MD, Department of Surgery, University of Nevada School of Medicine, 2040 West Charleston Boulevard, Suite 601, Las Vegas, Nevada 89102. Presented at the 42nd Annual Meeting of the Southwestern Surgical Congress, La Quinta, California, April 22-25, 1990.
630
THE AMERICAN JOURNAL OF SURGERY
ecognition that blood transfusion enhances graft survival in patients undergoing kidney transplantaR tion focused attention on the immunosuppressive effects of blood components. Analysis of patients undergoing operation for cancer has revealed varying results for transfusion effects in patients with tumors of different organ systems. However, substantial data exist to suggest very strongly that blood transfusion and its subsequent immunosuppression has a significant, deleterious effect on patients with early stage cancer undergoing surgical resection [1]. In particular, several reports of patients undergoing surgical resection for lung cancer exist [2-5]. Although they contain some conflicting results and conclusions, most have suggested a deleterious transfusion effect in this patient population. Several of these reports are difficult to interpret because of the inclusion of patients with a relatively advanced disease stage. Because the prognosis of these patients is worse than for patients with earlierstage disease, it is more difficult to show an effect of blood transfusion. To further analyze this important issue in thoracic surgical oncology, we reviewed a series of 117 patients with stage I (T1,N0,M0 or T2,N0,M0) nonsmall-cell lung cancer to provide additional data concerning the effect of blood transfusion in these patients. PATIENTS AND M E T H O D S The study includes 117 patients with stage I nonsmall-cell lung cancer who received operative therapy and in whom complete follow-up was available either until death or until completion of the study. These patients were operated on either at the University of Chicago Hospitals, Chicago, Illinois, or at University Medical Center, Las Vegas, Nevada. There were 60 male and 57 female patients, with 54 white, 60 black, and 3 Asian patients. Mean and median follow-up for all patients was 49.7 months and 47 months, respectively. We did not include patients who died during the perioperative period to allow examination of the effect of blood transfusion on long-term survival. Patients were divided into two groups depending upon their transfusion history. Patients who received administration of either whole blood or packed red blood cells within 30 days before or after operation constituted the transfusion group. The remaining patients formed the no transfusion group. Statistical comparisons of the two groups were performed with chi-square analysis or analysis of variance. Survival analysis was performed using the Statistical Analysis System and the Statistical Package for the Social Sciences.
VOLUME 160
DECEMBER 1990
EFFECT OF PERIOPERATIVE BLOOD TRANSFUSION ON CANCER
TABLE I
O1
PerloperaUve Transfusion In Stage I Non-Small-Cell Lung Cancer (mean 4- SD) Transfusion (n = 58)
No Transfusion (n = 59)
,9~ .8| .7~
p Value
Patient 61.3 4- 8.8 60.1 4- 9.6 0.4719 Age (yr) M 32, F 26 M 28, F 31 NS Sex White 31, Black 31 White 29, Black 29 NS Race T1 31, T2 27 T1 34, T2 25 NS Primary tumor Cell type 26 (45%) 37 (63%) 0.0793 Adenocarcinoma 25 (43%) 16 (27%) 0.1056 Squamous cell carcinoma 7 (12%) 6 (10%) 0,7438 Large cell carcinoma Operation 10 (17%) 8 (14%) 0.7675 Pneumonectomy Other 48 (83%) 51 (86%) 0.5810 578.0 4- 488.6 320.4 4- 198.8 0.0003* Blood loss (mL) Hematoorit 37.7 4- 5.2 39.8 4- 4.5 0.0248" Admission 34.5 4- 4.6 34.8 4- 4,9 0.7897 Discharge * Difference significant.
i
.64 .5| .4 4
NO T x T.
0
20
40
60
so
100 12o 140 16o 18o 200
Months
Figure 1. This figure contrasts the disease-free survival curve for the 58 patients who received transfusion (TX) to the survival curve of 59 patients who did not receive transfusions. The difference in 5-year disease-free survival, as described in the text, is significant with p = 0.0055.
TABLE
II
5 - Y e a r Disease-Free Survival (mean -I- 1 SE)
RESULTS There were 117 patients with histologically confirmed non-small-cell lung cancer. These patients were operated on between January 1, 1977, and December 31, 1986. Sixty-three patients had adenocarcinoma, 41 patients had squamous cell carcinoma, and 13 patients had large cell carcinoma. Eighteen patients had pneumonectomy as their operation and 99 patients had either lobectomy, segmentectomy, or wedge resection as their surgical treatment. Preoperative staging varied but, in addition to history and physical examination, included chest radiographs in all patients, computed tomographic scans of the chest and upper abdomen in all patients after 1981, mediastinoscopy in 64 patients, and a variety of radionuclide scans in many patients. Final pathologic staging documented stage I disease in all patients, most of whom had systematic removal of multiple mediastinal lymph node basins during their operation. A total of 58 patients received transfusion, constituting the perioperative transfusion group, and 59 patients received no transfusion. Table I compares the transfused with the non-transfused groups. The only significant differences are in the average operative blood loss and the admission hematocrit. Computation of survival functions was based on the actuarial method of Berkson and Gage [6] and was performed using the Statistical Package for the Social Sciences. The overall cumulative 5-year disease-free survival rate was 67%. A statistic was calculated from the survival scores using the algorithm of Lee and Desu [7]. In patients with transfusion, 5-year disease-free survival was 53% and in patients without transfusion it was 81% (p = 0.0055) (Figure 1). Table II shows the effect on survival of multiple variables when considered independently.
Cumulative Survival (%) All patients (n = 117) Transfusion (n = 58) No transfusion (n = 59) Male (n = 60) Female (n = 57) White (n --- 54) Black (n --- 60) Adenocarcinoma (n = 63) Squamous (n = 41) Pneumonectomy (n = 18) Other operations (n = 99)
67 53 81 59 74 63 70 76 81 67 67
4- 0.051 4- 0.073 4- 0.066 4- 0.080 4- 0.065 4- 0.073 4- 0.072 4- 0.068 4- 0.80 4- 0.111 4- 0.057
p Value
-0.0055" 0.1765 0.3323 0.0143" 0.1428
* Difference significant.
A parametric accelerated failure time or regression model was fitted to the survival time using the Statistical Analysis System. The covariates are patient age, patient race, patient sex, cell type, extent of operation (pneumonectomy versus lobectomy/segmentectomy), operative blood loss, admission hematocrit, discharge hematocrit, and the presence or absence of perioperative transfusion. The model indicates that the only covariate that significantly correlates with 5-year disease-free survival is the presence or absence of perioperative transfusion (p = 0.0278). Table III shows that transfusion is associated with a worse prognosis for all patient subgroups. Interestingly, this correlation is greater in women than in men, in black than in white patients, and in patients with squamous cell carcinoma than in patients with adenocarcinoma. In addition, this effect is not related to the quantity of blood transfused since survival in patients receiving one or two units of blood (50%) is similar to that in patients receiving three or more units (59%) (p = 0.3043).
THE AMERICAN JOURNAL OFSURGERY
V O L U M E 160
D E C E M B E R 1990
631
LITTLE ET AL
TABLE III 5-Year Disease-Free Survival In Patient Subgroups Transfusion (%) All patients Males Females White Black Adenocarcinoma Squamous cell carcinoma
Pneumonectomy Other operations Transfusion 1-2 units 3 or more
53 47 60 48 59 64 49 50 55
44444444. 4-
0.073 0.107 0.100 0.108 0.100 0.107 0.105 0.110 0.106
50 4- 0.100 59 4- 0.104
No Transfusion (%) 81 71 90 79 83 85 80 88 80
44444444. 4-
0.066 0.116 0.055 0.077 0.099 0.080 0.103 0.105 0.090
p Value 0.0055 0.0851 0.0265 0.0699 0.0344 0.1114 0.0617 0.1145 0.0348 0.3043
COMMENTS Transfusion of blood components clearly has an immunosuppressive effect. Exposing a patient to blood product antigens stimulates the immune system to some extent, and the early immunology literature documented and emphasized this effect [8]. However, experience with and analysis of patients undergoing renal transplantation have unequivocally shown that perioperative transfusion improves graft survival through a nonspecific immunosuppressive action [9,10]. This immunosuppressive effect is incompletely explicated, but the data show that transfusion increases the number and/or activity of suppressor T lymphocytes, decreases the number of natural killer ceils, induces the formation of anti-idiotypic antibodies, and impairs lymphocyte blastogenesis [11-14] With this documented suppressive effect of blood components on the immune system, it is not surprising that there is the potential for a deleterious effect in patients undergoing operative treatment for cancer. This effect seems to vary between cancers arising in different organs. In lung cancers, several reports exist with varying results [2-5]. Unfortunately, interpretation of these studies is complicated by their inclusion of patients with relatively advanced disease, specifically patients with stage II and stage III lung cancers. In these patients, whose overall prognosis is considerably poorer than patients with stage I disease, it is harder to identify a real but modest effect on prognosis due to blood transfusions. Three of the four series, however, did identify a statistically significant deleterious effect on prognosis in transfused patients with stage I and/or II non-small-cell lung cancer undergoing resectional therapy, For this reason, we focused completely on patients with early stage non-small-cell lung cancer. Although a retrospective analysis like ours does not prove a clinically significant effect, the definite statistical correlation does indicate this conclusion very strongly, particularly when the two patient groups are so similar. There is a statistically significant difference in the admission hematocrit level but it is unlikely that this difference, 37.7 4- 5.2 versus 39.8 4- 4.5, is clinically significant. Independent comparison of variables suggests that both 632
transfusion and cell type correlate with survival. However, only the presence or absence of perioperative transfusion is a significant covariate in the regression model. The suggestions that this effect is more pronounced in women than in men, in black than in white patients, and in patients with squamous cell carcinoma than those with adenocarcinoma are interesting but require future corroboration by other investigators. Given the findings of this study, which strengthen similar conclusions from earlier investigations, it is clear that perioperative and intraoperative care of patients undergoing surgical resection of stage I non-small-cell lung cancer should include concerns about the effect of transfusion. A patient should not undergo elective transfusion simply because of a moderately low hemoglobin or hematocrit level. The need for transfusion should be based solely on specific indications such as evidence of inadequate tissue oxygenation. Under most circumstances, the patient should be allowed to replenish red blood cell and hemoglobin levels endogenously, aided by oral iron supplementation if necessary. Reinfusion of autologous blood during surgery is probably inadvisable because of the risk of contamination with cancer cells.
REFERENCES 1. Wu HS, Little AG. Perioperative blood transfusions and cancer recurrence. J Clin Oncol 1988; 6: 1348-54. 2. Tartter PI. Burrows L. Kirschner P. Perioperative blood transfusion adversely affects prognosis after resection of stage I (subset N O ) non-oat cell lung cancer. J Thorac Cardiovasc Snrg 1984; 88: 659-22. 3. Hyman NH, Foster RS, DeMeules JE, Costanza MC. Blood transfusions and survival after lung cancer resection. Am J Surg 1985; 149: 502-7. 4. Keller SM, Groshen S, Martini N, Kaiser LR. Bloodtransfusion and lung cancer recurrence. Cancer 1988; 62: 606-10. 5. Moores DW, Piantadosi S, McKneally MF. Effect of perioperarive blood transfusion on outcome in patients with surgically resected lung cancer. Ann Thorac Surg 1989; 47: 346-51. 6. BerksonT, Gage R. Calculation of survivalrates for cancer. Proc Mayo Clin 1950; 25: 270-6. 7. Lee E, Desu M. A computer program for comparing k samples with right-censored data. Comp Prog Biomed 1972; 2: 315-21. 8. Medawar PE. Immunityto homologousskin II. The relationship between antigens on blood and skin. Br J Exp Pathol 1946; 27: 1521. 9. Opelz G, Sengor DPS, McKey MR, Terasaki PI. Effect of blood transfusions on subsequent kidney transplants. Trans Proc 1973; 5: 253-9. 10. Opelz G, Terasaki PI. Improvement of kidney graft survival with increased numbers of blood transfusions. N Engl J Med 1978; 299: 799-804. 11. Fischer E, Lenhard V, Seifect P, et al. Blood transfusioninduced suppression of cellular immunity in man. Hum Immunol 1980; 1: 187-94. 12. Smith MD, Williams JD, Coles GA, et al. The effect of blood transfusion on T-suppressor cells in renal dialysis patients. Transplant Proc 1981; 13: 181-3. 13. Kaplan J, Sarnaik S, Gitlin J, et al. Diminishedhelper/suppressor lymphocyte ratios and natural killer activity in recipients of repeated blood transfusion. Blood 1984; 64: 308-10. 14. Gasim P, Zoumbos NC, Young NS. Immunologic abnormalities in patients receiving multiple blood transfusions. Ann Intern Med 1984; 100: 173-7.
THE AMERICAN JOURNAL OF SURGERY VOLUME 160 DECEMBER 1990
EFFECT OF PERIOPERATIVE BLOOD TRANSFUSION ON CANCER
DISCUSSION William F. Sasser (St. Louis, MO): This effect is presumed to be due to immunosuppression. Tumor size and lymph node involvement are the most reliable prognostic indices in lung cancer, although blood transfusion should be considered a significant factor for recurrence and survival similar to an elevated carcinoembryonic antigen, the cancer cell type, the lowered serum protein, and the elevated white blood cell count. In the future, molecular studies will solve the theoretical problem. The practical solution now is to restrict and prevent blood transfusions. There were several questions that I would like to ask. In a report of colon resections, patients who received blood transfusions had an increase in postoperative infections; did you observe any increase in sepsis or infection in the patients who received transfusions? The perioperative transfusion effect was 30 days before and 30 days after surgery. If preoperatively your patient was so anemic that blood was required, would you delay surgery 30 days? One requirement for transfusion as stated in the manuscript was proper tissue oxygenation. What
was your criterion for this? Is there a role for autologous blood collection 2 weeks before the operation in patients with stage I lung cancer? Are there any data to indicate that the patients' stored blood is altered such that it has an immunosuppressive influence when reinfused? Alex G. Little (closing): We did not look at the rate or incidence of sepsis in these patients. Is 30 days reasonable? I came up with 30 days simply because that is what other series have used. I doubt I would want to delay surgery 30 days after a transfusion in a patient with cancer just because we know that is a dangerous time period. Investigators who have looked at the duration of the immunosuppressive effect have found that it peaks at approximately 2 weeks. Again, I believe we should only perform transfusions when there are concerns about tissue oxygenation and perfusion. And finally, I think reinfusing shed blood has theoretic possibilities; however, even in patients with limited-stage cancer, it has been shown that malignant cells can enter the pleural space during surgery.
THE AMERICAN JOURNAL OF SURGERY VOLUME160 DECEMBER 1990 633