CORRESPONDENCE
sulting mia.
role of renal insufficiency when present. I. H. KHAN, M.R.C.P. The author suggests that uriUniversity of Aberdeen nary losses of EPO in NS may not Aberdeen, Scotland necessarily cause a deficiency 1. Vaziri ND, Kaupke CJ. Barton CH. Gonzales E. state and that the urinary losses Plasma concentration and urinary excretion of may be fully compensated for by erythropoietin in adult nephrotic syndrome. Am J a concomitant increased syntheMed 1992; 92: 35-40. 2. Chamfer C, Garnett ES, Parsons V. Veal1 N. Gkrsis. While this can occur, is it not merular filtration rate measurement in man by the reasonable to assume at least a sin&-injection method using 51 Cr-EDTA. Clin Sci partial causality between abnorMol Med 1969; 37: 169-60. 3. Basu TK. Stein RM. Erythrocytosis associated mal losses and relative deficiency with chronic renal disease. Arch Intern Med 1974; when present simultaneously? 133: 442-7. The author has noted that 4. Myers DI. Ciuffo AA. Cooke CR. Focal $omerulomeasurement of GFR by radiosclerosis and erythrocytosis. Johns Hopkins Med J 1979; 145: 192-5. isotope technique would have 5. Chen YC, Yeh JC. Hsu HC. Secondary polycythe been more desirable than use of mia associated with membranous nephropathy. serum creatinine. We fully agree Clin Nephrol 1989; 33: 148-51. with this statement. However, isSubmitted February 20.1992, and accepted May 7. sues of cost and patient coopera1992 tion precluded the use of more reliable measurements of renal The Reply: function such as inulin clearance Dr. Khan has raised a number of or radioisotope techniques. issues regarding our recent article For obvious reasons it is often difficult to fully control for a published in The American Journal of Medicine in which we myriad of variables in a clinical study. With this in mind, we carreported altered EPO metabolism in NS. The author has stated ried out a series of careful animal that we had not excluded the role studies in which nephrotic-range of renal insufficiency. Clearly, proteinuria was produced in the mild to moderate renal insuffiabsence of renal insufficiency. The nephrotic rats in this study ciency, present in a subgroup of our patients, was in part responshowed a markedly attenuated sible for reduced EPO producEPO response to the induction of tion. However, we observed low graded anemia and hypoxemia EPO values in those with serum when compared with the normal creatinine levels below 1.5 controls. This was associated mg/dL. This was clearly exempliwith marked urinary excretion of fied by Patient 10, who had a nor- EPO, which greatly blunted the mal serum creatinine concentraEPO increase in response to anetion (1.1 mg/dL) and severe ane- mia and hypoxemia. This paper mia (hematocrit = 22%) but a was recently presented at the plasma EPO concentration (6 American Society of Nephrology mU/mL) below the mean value meeting (November 1991, Baltifound in the normal control more, Maryland) [l]. The findgroup (6.7 mU/mL). The author ings of this carefully controlled has further implied that we did laboratory study have strongly not acknowledge the contribuconfirmed the conclusions tion of renal insufficiency in di- reached in our clinical study. minished EPO response to aneWith regard to secondary mia. To the contrary, we expliciterythrocytosis, the very small ly noted the multifactorial nature number of reported cases (12 paof altered EPO metabolism in tients) relative to the huge pool of this population (page 38, Compatients with NS indicates its exments section), emphasizing the treme rareness. Renal ischemia 356
in secondary
polycythe-
September 1992 The American Journal of Medicine
Volume 93
due to volume contraction, OI even more likely intrarena thrombotic events, and possibly concurrent but unrelated phe. nomena rather than proteinuris per se may be responsible for the rare reported cases of secondary erythrocytosis. Incidentally, according to careful studies, NS ie not necessarily associated with intravascular contraction. On the contrary, plasma volume is usually increased, not decreased, in NS Dl. N.D. VAZZRZ,M.!., F.A.C.P. University of California, Irvine College of Medicine Orange, California 1. Zhou XJ, Vaziri ND. Erythropoietin metabolism and pharmacokinetia in rats with puromycin-aminonucleoside (PAN) induced nephrosis. J Am Sot Nephrol 1991; 2: 391. 2. Perico N. Delaini F, Lupini C, eta/. Blunted excretory response to atrial natriuretic peptide in experimental nephrosis. Kidney Int 1989; 36: 57-64.
ESTIMATESOF RISK OF POSTTRANSFUSION HEPATITIS To the Editor: Carson et al [l] provide valuable insights into the relative contributions of non-A, non-B hepatitis and human immunodeficiency virus (HIV) infection to transfusion-associated risk and the relatively small effect of blood transfusions on life expectancy. We offer reasons why transfusion may have even less impact on average life expectancy. The authors noted the importance of the anti-hepatitis C (HCV) assay, first used to test donated units in 1990, to declining risks of posttransfusion hepatitis. Also important to the declining risk of posttransfusion hepatitis, however, is the reduced prevalence of HCV seropositivity in blood donor pools (currently about 0.5% per donor) [2], likely the result of tighter donor screening criteria. Considering the antiHCV screening assay, current do-
nor profiles, and several recent reports, we estimate the incidence of non-A, non-B hepatitis to be approximately 0.1% per unit [2-51.
To model the sequelae of nonA, non-B hepatitis, the authors stratified patient outcomes by histopathology (chronic persistent hepatitis, chronic active hepatitis, and cirrhosis). They used natural history data from studies of non-transfusion-acquired hepatitis B. As the article points out, the literature does not allow adequate correlation between histopathologic state and clinical outcome in patients with posttransfusion hepatitis. Furthermore, transition rates between histopathologic states are poorly quantified. When we ignored histopathology and considered clinical outcome only in long-term (3 to 10 years) followup studies of patients with chronic posttransfusion hepatitis [6-8], we calculated the liver-related mortality rate to be 0.35% per year (4.5 liver-related deaths in about 1,250 patient-years of follow-up). The excess mortality rate of posttransfusion hepatitis derived this way is considerably lower than that assumed in the authors’ decision analysis model. Assuming a 0.3% per-unit risk of non-A, non-B hepatitis, Carson et al estimated that 5.7 days of life were lost when 2 units of blood were transfused to a 50year-old patient. We reconstructed the decision analysis model for a 50-year-old patient, using the authors’ assumptions of transfusion reactions and HIV infection and our assumptions of non-A, non-B hepatitis incidence and natural history. By this revised model, posttransfusion hepatitis still represents the large majority of transfusion-associated risk. However, our estimates reveal that only 1 day of life is lost by the transfusion of 2 units of blood to a 50-year-old patient.
As Carson et al conclude, efforts to improve the safety of blood should focus on reducing the risk of non-A, non-B hepatitis. Given the already low risks of transfusion, however, the net effect on public health of increasing blood safety may reflect the law of diminishing returns.
lar carcinoma. At the time of the preparation of our decision analysis, the field was rapidly changing with new studies coming out monthly. In fact, since the publication of our paper, a more sensitive second-generation hepatitis C assay has been adopted in blood banks across the United JOHN D. BIRKMEYER, M.D. States. JAMESP. AUBUCHON, M.D. Birkmeyer and colleagues sugBENJAMIN LITTENBERG,M.D. gest that our estimate of the rate Dartmouth-Hitchcock Medical Center of posttransfusion hepatitis C is Lebanon, New Hampshire too high (0.3% per unit) and 1. Carson JL. Russell LB, Taragin MI, Sonnenberg should be 0.1% per unit. Using FA, Duff AE, Bauer S. The risks of blood transfusion: their value in our analysis, the the relative influence of acquired immunodeficiency number of days lost for a 2-unit syndrome and non-A. non-6 hepatitis. Am J Med transfusion in a 50-year-old man 1992; 92: 45-52. 2. Dodd RY. Will blood products be free of infectious would be reduced from 5.7 to 2.0 agents? In: Nance SJ. editor. Transfusion medicine days. This is still much greater in the 1990’s. Arlington, VA: American Association than the days lost from acquired of Blood Banks, 1990: 223-51. immunodeficiency syndrome, 3. Van Der Poel CL, Reesink HW, Schaasbert W. et al. Infectivity of blood seropositive for hepatitis C which reemphasizes our concluvirus antibodies. Lancet 1990; 335: 55860. sion that most of the risk from a 4. Barrera JM, Bruguera M, Ercilla MG, et al. Inciblood transfusion is due to nondence of non-A. non-B hepatitis after screening blood donors for antibodies to hepatitis C virus and A, non-B hepatitis. surrogate markers. Ann Intern Med 1991; 115: However, the studies they 596400. quote are from Europe, where the 5. Contreras M. Barbara JAI, Anderson CC, et al. risk appears to be lower than in Low incidence of non-A. non-B post-transfusion hepatitis in London confirmed by hepatitis C virus serolthe U.S. [2-4]. Interestingly, a ogy. Lancet 1991; 337: 753-7. more recent report from the U.S. 6. Alter HJ. Transfusion-associated non-A, non-B suggests that the risk was 0.9% hepatitis: the first decade. In: Viral hepatitis and after testing with the first-generliver disease. New York: Alan R Liss. 1988: 537-42. 7. Koretz RL, Stone 0. Mousa M. GitnickGL. Non-A, ation hepatitis C assay, and 0.3% non-B posttransfusion hepatitis-a decade later. with the new second-generation Gastroenterology 1985; 88: 1251-4. assay now used in the U.S. [5]. 6. Realdi G, Termolada F, Bortolotti F, et al. The Thus, we believe our estimate of natural history of post-transfusion and sporadic non-A, non-B hepatitis in Italy. In: Viral hepatitis and the risk of hepatitis C is still apdelta infection. New York: Alan R Liss. 1983: 55-66. plicable. Submitted March 6. 1992, and accepted May 25. Birkmeyer et al were also con1992 cerned about our estimates of the natural history of hepatitis C. As The Reply: they point out, we acknowledge Remarkable progress has been that there is limited information made to improve the safety of ho- on the natural history of this dismologous blood transfusions. ease since the assay for hepatitis Only a decade ago, the risk of C has just become available. posttransfusion hepatitis was be- Their approach to estimating the tween 5% and 10% per unit. In natural history of posttransfu1989, Kuo and colleagues [l] de- sion hepatitis is reasonable, but scribed an assay to detect antino more strongly supported by bodies to hepatitis C. Since 1989, the data than our approach. a large number of reports have There are no large prospective sedocumented an association be- ries with well-documented and tween hepatitis C and posttranscarefully followed patients with fusion hepatitis and hepatocelluconfirmed hepatitis C. September 1992 The American Journal of Medicine
Volume 93
357
sulting mia.
role of renal insufficiency when present. I. H. KHAN, M.R.C.P. The author suggests that uriUniversity of Aberdeen nary losses of EPO in NS may not Aberdeen, Scotland necessarily cause a deficiency 1. Vaziri ND, Kaupke CJ. Barton CH. Gonzales E. state and that the urinary losses Plasma concentration and urinary excretion of may be fully compensated for by erythropoietin in adult nephrotic syndrome. Am J a concomitant increased syntheMed 1992; 92: 35-40. 2. Chamfer C, Garnett ES, Parsons V. Veal1 N. Gkrsis. While this can occur, is it not merular filtration rate measurement in man by the reasonable to assume at least a sin&-injection method using 51 Cr-EDTA. Clin Sci partial causality between abnorMol Med 1969; 37: 169-60. 3. Basu TK. Stein RM. Erythrocytosis associated mal losses and relative deficiency with chronic renal disease. Arch Intern Med 1974; when present simultaneously? 133: 442-7. The author has noted that 4. Myers DI. Ciuffo AA. Cooke CR. Focal $omerulomeasurement of GFR by radiosclerosis and erythrocytosis. Johns Hopkins Med J 1979; 145: 192-5. isotope technique would have 5. Chen YC, Yeh JC. Hsu HC. Secondary polycythe been more desirable than use of mia associated with membranous nephropathy. serum creatinine. We fully agree Clin Nephrol 1989; 33: 148-51. with this statement. However, isSubmitted February 20.1992, and accepted May 7. sues of cost and patient coopera1992 tion precluded the use of more reliable measurements of renal The Reply: function such as inulin clearance Dr. Khan has raised a number of or radioisotope techniques. issues regarding our recent article For obvious reasons it is often difficult to fully control for a published in The American Journal of Medicine in which we myriad of variables in a clinical study. With this in mind, we carreported altered EPO metabolism in NS. The author has stated ried out a series of careful animal that we had not excluded the role studies in which nephrotic-range of renal insufficiency. Clearly, proteinuria was produced in the mild to moderate renal insuffiabsence of renal insufficiency. The nephrotic rats in this study ciency, present in a subgroup of our patients, was in part responshowed a markedly attenuated sible for reduced EPO producEPO response to the induction of tion. However, we observed low graded anemia and hypoxemia EPO values in those with serum when compared with the normal creatinine levels below 1.5 controls. This was associated mg/dL. This was clearly exempliwith marked urinary excretion of fied by Patient 10, who had a nor- EPO, which greatly blunted the mal serum creatinine concentraEPO increase in response to anetion (1.1 mg/dL) and severe ane- mia and hypoxemia. This paper mia (hematocrit = 22%) but a was recently presented at the plasma EPO concentration (6 American Society of Nephrology mU/mL) below the mean value meeting (November 1991, Baltifound in the normal control more, Maryland) [l]. The findgroup (6.7 mU/mL). The author ings of this carefully controlled has further implied that we did laboratory study have strongly not acknowledge the contribuconfirmed the conclusions tion of renal insufficiency in di- reached in our clinical study. minished EPO response to aneWith regard to secondary mia. To the contrary, we expliciterythrocytosis, the very small ly noted the multifactorial nature number of reported cases (12 paof altered EPO metabolism in tients) relative to the huge pool of this population (page 38, Compatients with NS indicates its exments section), emphasizing the treme rareness. Renal ischemia 356
in secondary
polycythe-
September 1992 The American Journal of Medicine
Volume 93
due to volume contraction, OI even more likely intrarena thrombotic events, and possibly concurrent but unrelated phe. nomena rather than proteinuris per se may be responsible for the rare reported cases of secondary erythrocytosis. Incidentally, according to careful studies, NS ie not necessarily associated with intravascular contraction. On the contrary, plasma volume is usually increased, not decreased, in NS Dl. N.D. VAZZRZ,M.!., F.A.C.P. University of California, Irvine College of Medicine Orange, California 1. Zhou XJ, Vaziri ND. Erythropoietin metabolism and pharmacokinetia in rats with puromycin-aminonucleoside (PAN) induced nephrosis. J Am Sot Nephrol 1991; 2: 391. 2. Perico N. Delaini F, Lupini C, eta/. Blunted excretory response to atrial natriuretic peptide in experimental nephrosis. Kidney Int 1989; 36: 57-64.
ESTIMATESOF RISK OF POSTTRANSFUSION HEPATITIS To the Editor: Carson et al [l] provide valuable insights into the relative contributions of non-A, non-B hepatitis and human immunodeficiency virus (HIV) infection to transfusion-associated risk and the relatively small effect of blood transfusions on life expectancy. We offer reasons why transfusion may have even less impact on average life expectancy. The authors noted the importance of the anti-hepatitis C (HCV) assay, first used to test donated units in 1990, to declining risks of posttransfusion hepatitis. Also important to the declining risk of posttransfusion hepatitis, however, is the reduced prevalence of HCV seropositivity in blood donor pools (currently about 0.5% per donor) [2], likely the result of tighter donor screening criteria. Considering the antiHCV screening assay, current do-
nor profiles, and several recent reports, we estimate the incidence of non-A, non-B hepatitis to be approximately 0.1% per unit [2-51.
To model the sequelae of nonA, non-B hepatitis, the authors stratified patient outcomes by histopathology (chronic persistent hepatitis, chronic active hepatitis, and cirrhosis). They used natural history data from studies of non-transfusion-acquired hepatitis B. As the article points out, the literature does not allow adequate correlation between histopathologic state and clinical outcome in patients with posttransfusion hepatitis. Furthermore, transition rates between histopathologic states are poorly quantified. When we ignored histopathology and considered clinical outcome only in long-term (3 to 10 years) followup studies of patients with chronic posttransfusion hepatitis [6-8], we calculated the liver-related mortality rate to be 0.35% per year (4.5 liver-related deaths in about 1,250 patient-years of follow-up). The excess mortality rate of posttransfusion hepatitis derived this way is considerably lower than that assumed in the authors’ decision analysis model. Assuming a 0.3% per-unit risk of non-A, non-B hepatitis, Carson et al estimated that 5.7 days of life were lost when 2 units of blood were transfused to a 50year-old patient. We reconstructed the decision analysis model for a 50-year-old patient, using the authors’ assumptions of transfusion reactions and HIV infection and our assumptions of non-A, non-B hepatitis incidence and natural history. By this revised model, posttransfusion hepatitis still represents the large majority of transfusion-associated risk. However, our estimates reveal that only 1 day of life is lost by the transfusion of 2 units of blood to a 50-year-old patient.
As Carson et al conclude, efforts to improve the safety of blood should focus on reducing the risk of non-A, non-B hepatitis. Given the already low risks of transfusion, however, the net effect on public health of increasing blood safety may reflect the law of diminishing returns.
lar carcinoma. At the time of the preparation of our decision analysis, the field was rapidly changing with new studies coming out monthly. In fact, since the publication of our paper, a more sensitive second-generation hepatitis C assay has been adopted in blood banks across the United JOHN D. BIRKMEYER, M.D. States. JAMESP. AUBUCHON, M.D. Birkmeyer and colleagues sugBENJAMIN LITTENBERG,M.D. gest that our estimate of the rate Dartmouth-Hitchcock Medical Center of posttransfusion hepatitis C is Lebanon, New Hampshire too high (0.3% per unit) and 1. Carson JL. Russell LB, Taragin MI, Sonnenberg should be 0.1% per unit. Using FA, Duff AE, Bauer S. The risks of blood transfusion: their value in our analysis, the the relative influence of acquired immunodeficiency number of days lost for a 2-unit syndrome and non-A. non-6 hepatitis. Am J Med transfusion in a 50-year-old man 1992; 92: 45-52. 2. Dodd RY. Will blood products be free of infectious would be reduced from 5.7 to 2.0 agents? In: Nance SJ. editor. Transfusion medicine days. This is still much greater in the 1990’s. Arlington, VA: American Association than the days lost from acquired of Blood Banks, 1990: 223-51. immunodeficiency syndrome, 3. Van Der Poel CL, Reesink HW, Schaasbert W. et al. Infectivity of blood seropositive for hepatitis C which reemphasizes our concluvirus antibodies. Lancet 1990; 335: 55860. sion that most of the risk from a 4. Barrera JM, Bruguera M, Ercilla MG, et al. Inciblood transfusion is due to nondence of non-A. non-B hepatitis after screening blood donors for antibodies to hepatitis C virus and A, non-B hepatitis. surrogate markers. Ann Intern Med 1991; 115: However, the studies they 596400. quote are from Europe, where the 5. Contreras M. Barbara JAI, Anderson CC, et al. risk appears to be lower than in Low incidence of non-A. non-B post-transfusion hepatitis in London confirmed by hepatitis C virus serolthe U.S. [2-4]. Interestingly, a ogy. Lancet 1991; 337: 753-7. more recent report from the U.S. 6. Alter HJ. Transfusion-associated non-A, non-B suggests that the risk was 0.9% hepatitis: the first decade. In: Viral hepatitis and after testing with the first-generliver disease. New York: Alan R Liss. 1988: 537-42. 7. Koretz RL, Stone 0. Mousa M. GitnickGL. Non-A, ation hepatitis C assay, and 0.3% non-B posttransfusion hepatitis-a decade later. with the new second-generation Gastroenterology 1985; 88: 1251-4. assay now used in the U.S. [5]. 6. Realdi G, Termolada F, Bortolotti F, et al. The Thus, we believe our estimate of natural history of post-transfusion and sporadic non-A, non-B hepatitis in Italy. In: Viral hepatitis and the risk of hepatitis C is still apdelta infection. New York: Alan R Liss. 1983: 55-66. plicable. Submitted March 6. 1992, and accepted May 25. Birkmeyer et al were also con1992 cerned about our estimates of the natural history of hepatitis C. As The Reply: they point out, we acknowledge Remarkable progress has been that there is limited information made to improve the safety of ho- on the natural history of this dismologous blood transfusions. ease since the assay for hepatitis Only a decade ago, the risk of C has just become available. posttransfusion hepatitis was be- Their approach to estimating the tween 5% and 10% per unit. In natural history of posttransfu1989, Kuo and colleagues [l] de- sion hepatitis is reasonable, but scribed an assay to detect antino more strongly supported by bodies to hepatitis C. Since 1989, the data than our approach. a large number of reports have There are no large prospective sedocumented an association be- ries with well-documented and tween hepatitis C and posttranscarefully followed patients with fusion hepatitis and hepatocelluconfirmed hepatitis C. September 1992 The American Journal of Medicine
Volume 93
357
