Vox Sang. 44: 212-217 (1983) 0 1983 S . Karger AG. Basel

0042-9007/83/0444-0212 $2.7510

Immune Response to Chronic Red Blood Cell Transfusion

Neil Blumberg, Kathy Peck, Karen Ross, Eduardo Avila Blood Bank, Clinical Pathology Laboratories and Department of Pathology, University of Rochester Medical Center, Rochester, N.Y., USA

Abstract. Patients receiving chronic transfusion for aplastic anemia or hemoglobinopa- thy are believed to be at high risk for developing red blood cell alloantibodies, while those undergoing chemotherapy for leukemia are believed to be at low risk. To test this hypothesis, we studied the acquisition ofnew alloantibodies in 703 transfused patients. While none of 99 patients with lymphocytic leukemia made new antibodies, patients with myelogenous leu- kemia (l6%), hemoglobinopathy (29%), aplastic anemia (llo/o), gastrointestinal bleeding (11%) or renal failure (14%) made antibodies at statistically similar rates. Lymphocytic leukemia or its treatment is characterized by a lack of immunologic response to transfusion. Patients with hemoglobinopathy or aplastic anemia do not appear at statistically significant greater risk of alloimmunization than many other patients requiring chronic transfusion. Neither intensive chemotherapy for myelogenous leukemia nor the uremia of renal failure significantly suppress the formation of blood group antibodies.

Introduction

Chronic red blood cell transfusion for anemia is both the oldest and most widely employed organ transplant. However, de- spite the ready availability of techniques to study the immune response to red blood cell antigens, little is known about the variation in recipient response in the different disease states in which this organ transplant has become indispensable [l-81. Variation in likelihood ofacquiring new antibodies to red blood cell antigens is ofclinical and econom- ic consequence. Development of such anti- bodies may compromise the care of chroni-

cally transfused patients. Bone marrow transplantation is becoming a common ther- apy for aplastic anemia and leukemias, and the presence of blood group antibodies creates technical complications, and con- tributes to morbidity or mortality. Because some patients with transfusion-dependent hemoglobinopathies become difficult to transfuse due to multiple blood group anti- bodies, prophylactic use of blood matched for multiple antigens other than ABO and Rho(D) has been recommended in some clinical situations [6]. Similar potential problems, while heretofore unexamined, may attend those patients with chronic renal

Immune Response to Red Cell Transfusion 213

failure and chronic gastrointestinal blood loss who need repeated or emergency trans- fusions of red blood cells. Since the cost of providing red blood cells matched for mul- tiple antigens may be several times that for transfusions matched only for ABO and Rho(D), there are economic considerations to this issue.

We retrospectively studied 703 patients with conditions commonly requiring multi- ple red blood cell transfusions to determine ifthe rate ofblood group antibody formation varied amongst these disparate diseases.

Methods

The names and hospital-identifying numbers of all patients with discharge diagnoses of leukemia (all types), aplastic anemia, gastrointestinal bleeding (all types), chronic renal failure and hemoglobinopathy (all types) during 1970-1980 at our institution were ob- tained by searching a computer data base. Patients with diagnoses of aplastic anemia secondary to leukemia were included in the latter group. Due to the large number of patients with diagnoses of gastrointestinal bleeding or chronic renal failure, only those patients discharged from 1978 to 1980 were studied. Records of transfusions and immunohematological studies on these patients were obtained from blood bank files. These included transfusions and tests for each patient from all previous admissions as well. Information col- lected includedage, sex, transfusion history (date, quan- tity and type of blood product), and results of tests for red blood cell alloantibodies.

During the period 1970-1980 a variety of immuno- hematological screening tests were employed. Common to all was a test of recipient serum against two reagent red cell preparations of known phenotype using anti- globulin technique (ix., the indirect Coombs’ test). The period of time, in weeks, from the first red blood cell transfusion in our institution to the last recorded indi- rect antiglobulin test was considered the ‘follow-up’ period and recorded. Patients who received no transfu- sions and/or had follow-up periods of less than 6 weeks are excluded from the results reported. When an anti- body had been detected, we recorded its specificity, the number of previous red cell transfusions and the num-

ber of weeks since the first transfusion. Antibodies pres- ent prior to the first transfusion were not included in the analysis.

Antibodies believed to occur ‘naturally’ without transfusion ( e g , Lea, Leb, M, N, P, , A , , etc.) were included in the analysis when they were not present prior to the first transfusion, but autoantibodies (e.g., anti-l) were excluded from the analysis. In any case, when such ‘naturally occurring’ antibodies were ex- cluded from the analysis, the results were identical.

In general, all patients received red blood cell con- taining transfusions compatible with their ABO and Rho(D) phenotype. These included packed red cells, whole blood, washed and frozen deglycerolyzed red cells. A group of 14 patients with hemoglobinopathy who received blood matched for additional antigens were excluded from the analysis (8 had thalassemia major and 6 had sickle cell anemia). 2 ofthese patients made alloantibodies (anti-Lea and anti-Yka) and 2 developed cold autoimmune hemolytic disease with anti-I in their serum. 18 of 240 patients with leukemia who could not be classified as having myeloid or lym- phoid variants are not included in the data presented. The numbers of patients with chronic myeloid or lym- phocytic leukemia were small (29 and 24, respectively), and the proportion with antibodies was similar to that in the acute forms ofthese diseases. Thus patients with chronic leukemias are grouped with those with the cor- responding acute forms ofthe disease in the analysis. Of the 52 patients with hemoglobinopathy, 2 had thalas- semia major and 50 had sickle cell anemia, or one of its variants.

Statistical analysis ofthe proportion of patients with and without new antibodies in the different diseases was performed using xz for 2 x2 contingency tables employ- ing Yates’ correction. Additionally, statistical signifi- cance was corrected using Bonferroni’s procedure for multiple comparisons. That is, since 14 comparisons were made, the usually required p value of 0.05 was divided by 14. Statistical significance thus required a p <0.0036. These stringent statistical requirements were employed in view of the uncontrolled, retrospec- tive nature of the data and the lack of comparability amongst the disease groups as to age, number of trans- fusions and follow-up periods. We recognized and ac- cepted that this strategy increases the possibility of a type11 (p) error. However, use ofx2 for 2x6 contingency tables and decomposition by 2 x 5 and 2 x4 arrays con- firmed that the major contribution to variation was due to the lymphocytic leukemia group [9].

2 I4 Blumberg/Peck/Ross/Avila

Table 1. Transfusion and follow-up data

Disease group Number of patients Number of transfusions per patient

Weeks of follow-up per patient

mean range

Gastrointestinal bleeding I98 23 1-256 Renal failure I78 47 3-414 Hemoglobinopathy 52 40 3-184 Myeloid leukemia 123 23 4-122 Lymphocytic leukemia 99 13 1-155 Aplastic anemia 53 41 1-299

mean range

262 6-1.340 140 6-966 284 6-1,261 44 6 4 6 1 63 6 4 7 0 71 6-528

Table 11. Alloimmunization rates amongst the disease groups

Disease group Number of patients Patients with Patients with multiple new antibodies new antibodies

n Yo n %O

Gastrointestinal bleeding I98 Renal failure I78 Hemoglobinopathy 52 Myeloid leukemia I23 Lymphocytic leukemia 99 Aplastic anemia 53

21 1 1 9 5 25 14 I I 6 15 29 4 8 20 16 5 4 0 0 0 0 6 I I 5 9

Results

The mean number of transfusions and weeks of follow-up and the total number of patients in each disease group are shown in tableI. Because the data on numbers of transfusions and follow-up periods are not normally distributed, ranges rather than standard deviations are reported for these data. The incidence of patients with new antibodies formed in response to transfusion is shown in table 11. The proportion of pa- tients making new blood group antibodies was not statistically significantly different in the varied diseases except: (1) patients with

hemoglobinopathies made more antibodies than those with gastrointestinal bleeding (p<0.002) and (2) patients with lympho- cytic leukemia made fewer antibodies than those with gastrointestinal bleeding (p < 0.002), renal failure (p< 0.002), hemo- globinopathy (p < 9 x lo-'), myeloid leu- kemia (p < 7 x or aplastic anemia (p < 0.003). The proportion of patients mak- ing multiple new antibodies was the same in all groups.

To determine if the formation of red blood cell alloantibodies was influenced by the number of transfusions given, we exam- ined the rate of antibody formation at early,

Immune Response to Red Cell Transfusion 215

middle, and late points in the course oftrans- fusion. Antibody formation occurred early in the transfusion course rather than later for all disease groups. While the mean number oftransfusions overall was 30,55% (48 of87) ofpatients had made their first new antibody at or prior to their 10th transfusion. The number of transfusions given overall per alloimmunized patient was: gastrointestinal bleeding (217), renal failure (333, hemoglo- binopathy (139), myeloid leukemia (141), and aplastic anemia (362) .

The specificities of the antibodies formed were limited. If the ‘naturally occurring’ an- tibodies are excluded, fully 90% of the anti- bodies were in the Rh and Kell blood groups. Except for slightly higher numbers of Lewis antibodies in the hemoglobinopathy pa- tients and of anti-N in the chronic renal fail- ure group, no differences existed in the spec- ificities of antibodies formed among the dis- ease groups.

Discussion

Within the limits of the retrospective methodology employed, information con- cerning the immune response to red blood cell transfusion in different disease states is obtainable. Uncontrolled variables include previous transfusion and pregnancy, and transfusion given at other facilities. How- ever, it seems unlikely that patients with these diseases should differ significantly in their likelihood of being transfused else- where during the period they were receiving care in our institution. The elimination of antibodies detectable prior to the first trans- fusion should minimize the likelihood of detecting antibodies due to previous transfu- sion or pregnancy. Patients in the hemoglo-

binopathy and acute lymphocytic leukemia group were much younger and therefore less likely to have been pregnant. Yet the former made antibodies at equivalent rates to other patients and the latter did not. Thus it is probable that the decreased formation of antibodies in patients with lymphocytic leu- kemia is due to the pathophysiology of the disease, or its therapy, rather than to the absence of previous pregnancy or transfu- sion. There are data supporting reduced im- mune responsiveness in lymphatic malig- nancies [8, 10-121.

Other uncontrolled variables are the number of transfusions and length of follow- up period, which vary substantially among the disease groups. Our data concerning the early appearance of most antibodies and the fact that the primary immune response to red blood cell antigens is usually detectable within 3-6 weeks of transfusion lead us to believe these biases are not significant con- straints on interpreting the data.

Because the number of patients with he- moglobinopathy in our study is relatively small, it is not possible to definitively answer the question of whether their alloimmuniza- tion rate is higher than that of other chron- ically transfused patients. While there is a trend toward more frequent antibody forma- tion in this group, it is informative to note that if the number of patients with hemoglo- binopathy in our study were doubled, and the proportion making antibodies remained the same, there would still be no statistically significant difference in alloimmunization rates between this group and those with myeloid leukemias or aplastic anemia. Thus our data suggest that our patients with he- moglobinopathy are not significantly more likely to make red cell alloantibodies than patients with myeloid leukemia or aplastic

216 Blumberg/Peck/Ross/Avila

anemia, but may be more likely to do so than patients with gastrointestinal bleeding or renal failure.

The trend toward somewhat fewer pa- tients making antibodies in the gastrointes- tinal bleeding and renal failure groups only achieved statistical significance in the former case. These quantitative differences, if real, seem to readily fit in with theoretical explanations. Patients with gastrointestinal blood loss retain less of the transfused blood, and therefore may be exposed to less of an antigenic stimulus for a shorter period of time than non-bleeding patients. It is also likely that this group is the most heteroge- neous of those studied in age, sex and pri- mary disease. Their transfusion pattern is also likely typified by acute massive rather than chronic small transfusions. Patients with chronic renal failure are thought not to be fully immunocompetent [13, 141. How- ever, our data suggest that their ability to make red cell antibodies is essentially intact.

Despite previously reported evidence of the immunosuppressive effects of therapy [15-171, our patients receiving chemother- apy for myeloid leukemia were not immu- nosuppressed with regard to their ability to make blood group antibodies. There is, in fact, ancillary evidence that chemotherapy might not greatly reduce antibody response to transfused blood cells. Refractoriness to platelet and granulocyte transfusion, pre- sumably immunologic in nature, remains a problem in the care of patients receiving intensive chemotherapy [18].

Whereas our data cannot completely ex- clude the possibility of modest quantitative differences in alloimmunization rates among chronically transfused patients, pa- tients with lymphocytic leukemia appear strikingly different from those in the other

disease groups. Not a single patient with lymphocytic leukemia made a new blood group antibody over the last 10 years at our institution. Although this group tended to receive fewer transfusions and be younger than those in other disease groups, we think it improbable that these two variables ex- plain our findings. In the patients making antibodies, they occurred very early in the transfusion course. Many patients in the hemoglobinopathy group, where the highest incidence of antibody formation was ob- served, were also young. None ofthe chronic lymphocytic leukemia patients, who are in general older, made new blood group alloan- tibodies. The two most likely contributions to this finding are the disease itself, or its chemotherapy. Because immune respon- siveness is probably aberrant in lymphoid malignancies [8, 10-121, we would speculate that the disease pathophysiology is the pre- dominant contributor, rather than the che- mot herapeutic agents employed.

Our findings conflict with some aspects of general clinical opinion, and represent the initial attempt to obtain this type ofcompar- ative data. Independent confirmation of our findings is necessary prior to drawing firm conclusions. A prospective study would likely clarify the effects of some of the un- controlled variables we have described. We will address the implications of our data for antigen-matching and its cost-effectiveness in a subsequent communication.

Acknowledgements

We are grateful to Drs. DArvan, C.Chuang, H . Cohen, J . Heal. J . Nushacher. E. Snyder and Ms. A . McMican for their thoughtful discussions. Mrs. Re- becca Cudreguri provided exemplary secretarial assis- tance.

Immune Response to Red Cell Transfusion 217

References

1 Wallace, J. M.; Henry, J . B.: lsoimmunization after massive transfusion for open heart surgery. Trans- fusion 5: 153-157 (1965).

2 Lostumbo, J.J.; Holland, P.V.; Schmidt, P.J.: Isoimmunization after multiple transfusions. New Engl. J. Med. 275: 141-144 (1966).

3 Economidou, J. ; Constantoulakis, M.; Augoustaki, 0.; Adinolfi, M.: Frequency ofantibodies to various antigenic determinants in polytransfused patients with homozygous thalassaemia in Greece. Vox Sang 20: 252-258 (1971).

4 Blumberg, B.S.; Alter, H.J.; Riddell, N. M.; Er- landson, M.: Multiple antigenic specificities of se- rum lipoproteins detected with sera of transfused patients. Vox Sang 9: 128-145 (1964).

5 Mollison, P. L.: Blood transfusion in clinical medi- cine; 6th ed., pp.202-203,212 (Blackwell, Oxford 1979).

6 Orlina, A.R.; Unger, P. J . ; Koshy, M.: Post-trans- fusion alloimmunization in patients with sickle cell disease. Am. J . Hemat. 5: 101-106 (1978).

7 Coles, S. M.; Klein, H.G.; Holland, P. V.: Alloim- munization in two multitransfused patient popula- tions. Transfusion 21: 462-466 (1981).

8 Fairley, G.H.; Akers, R.J.: Antibodies to blood groupA and B substances in reticuloses. Br. J. Hae- mat. 8: 375-391 (1962).

9 Snedecor, G. W.; Cochran, W.G.: Statistical meth- ods; 7th ed., pp.115-116. 124-128. 166. 208-213 (Iowa State University Press, Ames 1980).

10 Dupuy, J.M.; Kourilsky, F.M.; Fradelizzi, D.; Feingold, N.; Jacquillat, C.; Bernard, J . ; Dausset, J.: Depression of immunologic reactivity of patients with acute leukemia. Cancer 27: 323-331 (1971).

I I Han, T.; Ozer, H.; Henderson, E.S.; Dadey, B.; Nussbaum-Blumenson, A.; BdrCOS, M.: Defective

I

immunoregulatory T-cell function in chronic lym- phocytic leukemia. Blood 58: 1182-1188 (1981).

12 Shimizu, K.; Kitoh. S.: Increase in the number of lymphocytes secreting IgG following blood transfu- sion. Blood 5Y: 312-316 (1982).

13 Fehrman, 1 . ; Ringden, 0.: Lymphocytes from mul- titransfused uremic patients have poor MLC reac- tivity. Tissue Antigens 17: 386-395 (1981).

14 Kerman, R.H.: Floyd, M.; Van Buren, C.; Kahan, B. D.: Serial measurement of nonspecific immune parameters in chronically hemodialyzed renal fail- ure patients. J . clin. Immunol. I : 163-168 (1981).

15 Griswold. D. E.; Heppner. G. H.; Calabresi, P.: Se- lective suppression ofhumoral and cellular immun- ity with cytosine arabinoside. Cancer Res. 32:

16 Leventhal, B.G.; Cohen, P.; Triem. S.C.: Effect of chemotherapy on the immune response in acute leukemia. Israel J. med. Scis 10: 866-877 (1974).

17 Holohan, T. V.; Terasaki. P. 1.; Deisseroth, A. B.: Suppression of transfusion-related alloimmuniza- tion in intensively treated cancer patients. Blood 58:

I8 Schiffer, C. A. ; Aisner, J . ; Daly, P. A,; Schimpff, S.; Wiernik, P. H.: Alloimmunization following pro- phylactic granulocyte transfusion. Blood 54:

298-301 (1972).

122-127 (1981).

766-774 (1979).

Received: April 25, 1982 Accepted: August 10, 1982

Dr. Neil Bluniberg, Box 608, University of Rochester Medical Center, 601 Elmwood Avenue, Rochester, NY 14642 (USA)