Letters

Letters

The Editors welcome submissions for possible publication in the Letters section. Authors of letters should: • Include no more than 300 words of text, three authors, and five references • Type with double-spacing • Send three copies of the letter, a transfer-of-copyright form (see Table of Contents for location) signed by all authors, and a covering letter describing any conflicts of interest related to the contents of the letter • Provide a self-addressed envelope if they want to be notified that the letter was received Letters commenting on an Annals article will be considered if they are received within 6 weeks of the time the article was published. Only some of the letters received can be published. Published letters are edited and may be shortened; tables and figures are included only selectively. Authors will be notified that the letter has been received. If the letter is selected for publication, the author will be notified about 3 weeks before the publication date. Unpublished letters cannot be returned. Restriction Isotyping of Apolipoprotein E To the Editor: I read with interest the review by Walden and Hegele (1). The authors state that "apolipoprotein [apo] E is divided into two independently folded domains with different functions." The boundary between these two domains is the random-coil region spanning amino acids 165 to 201, which is subject to enzymatic cleavage (2). Thus, the region encompassing approximately the first 160 to 210 amino acids contains the apo B-E receptor-binding domain (which is located approximately between amino acids 130 and 160), whereas the C-terminal region mediates the binding to lipids. Evidence now suggests that apolipoproteins do specifically bind thyroid hormones with medium affinity (3-5). Such a hormone interaction may serve several biological functions (4). Studies of apo A-I and apo E have shown that the location is TV-terminal and that it coincides with the respective exon 3-coded region of both apolipoproteins. Sequences of particular relevance to this hormone-binding function are those having the predicted secondary structure of a /3-sheet (amino acids 32 to 44 in apo E and amino acids 9 to 23, 28 to 33, and 43 to 47 in apo A-I). By contrast, the secondary structure of the lipid-binding region is the amphipathic a-helix in the form of a 22-mer repeating unit. This location of the hormone site of apolipoproteins is supported by the recent finding that the exon 3-coded regions not only of apo A-I and apo E but also of the other apolipoproteins of the high-density lipoproteins share sequence homology with the regions of the three major thyroid hormone plasma transport proteins (thyroid-binding globulin, transthyretin, and albumin) known or postulated to contain the respective thyroid hormone sites (5). The TV-terminal position of the thyroid hormone site explains why the amino acid mutations at position 112 (apo E4) or 158 (E2) leave intact the thyroid hormone-binding property. It can be concluded that the domain constituted by the first 160 to 210 residues is composed of three subdomains: the hormone-binding subdomain (approximately the first 61 amino acids), the apo B-E receptor-binding subdomain (approximately amino acids 130 to 160) and a lipid-binding subdomain between them. It seems more logical (and more justified on genetic grounds), however, to consider apo E (and the other high-density lipoprotein apolipoproteins) as comprising two distinct domains: the exon 3-coded domain and the exon-4 coded domain. The first represents the thyroid hormone-binding domain, and the second consists of the lipid-binding domain (amino acids 62 to 68

299 in apo E). In apo E, however, this domain evolved to contain a specialized region that interacts with the apo B-E receptor and that can be viewed as a subdomain separating two flanking lipid-binding subdomains. Salvatore Benvenga, MD National Institute of Diabetes and Digestive and Kidney Diseases National Institutes of Health Bethesda, MD 20892 References 1. Walden CC, Hegele RA. Apolipoprotein E in hyperlipidemia. Ann Intern Med. 1994;120:1026-36. 2. Wetterau JR, Aggerbeck LP, Rail SC Jr, Weisgraber KH. Human apolipoprotein E3 in aqueous solution. I. Evidence for two structural domains. J Biol Chem. 1988;263:6240-8. 3. Benvenga S, Cahnmann HJ, Robbins J. Characterization of thyroid hormone binding to apolipoprotein E: localization of the binding site in the exon 3-coded domain. Endocrinology. 1993;133:1300-5. 4. Benvenga S, Robbins J. Lipoprotein-thyroid hormone interactions. Trends Endocrinol Metab. 1993;4:194-8. 5. Benvenga S, Cahnmann HJ, Rader D, Kindt M, Facchiano A, Robbins J. Thyroid hormone binding to isolated human apolipoproteins A-II, C-I, C-II and C-III: homology in thyroxine binding sites. Thyroid. 1994;4:261-7.

In response: Dr. Benvenga provides evidence to support binding of thyroid hormone to a region of apo E. This observation is interesting, especially given the clinical association between thyroid hormone status and lipoprotein metabolism. Others have shown that thyroid hormone acts at the molecular level to reduce hepatic production of apo E (1) and to up-regulate hepatic low-density lipoprotein receptors (2), for which apo E is a ligand. More studies are required to determine whether any of these mechanisms can explain the association between hypothyroidism and hyperlipidemia, particularly type III hyperlipoproteinemia. Apolipoprotein E has recently been shown to bind /3-amyloid peptide, a major constituent of the senile plaques found in the brains of patients with Alzheimer disease (3). Moreover, the E4 isoform of apo E, which has a frequency of about 15% in whites, interacts with /3-amyloid much more avidly than does the common E3 isoform. The higher binding affinity of E4 for j3-amyloid and the capacity of j3-amyloid and E4 to form a complex latticework of monofibrillar structures in vitro provides a mechanism for the well-established statistical association between the E4 isoform and both familial and sporadic Alzheimer disease (4). Also, the amount of amyloid in the brains of patients with Alzheimer disease was found to be highest in patients homozygous for E4, who also experience the earliest clinical manifestations of the disease. Finally, apo E was found to have an isoformspecific effect on neuronal growth in dorsal-root ganglia (5), with the E3 isoform stimulating and the E4 isoform impairing neurite outgrowth and neuronal elongation. Recent biochemical, genetic, morphologic, and pathologic findings lend compelling evidence to support an important role for the E4 isoform of apo E in the pathogenesis of Alzheimer disease. Although these results are consistent and exciting, no prospective evidence suggests that apo E genotyping can be used clinically to identify persons at risk for Alzheimer disease. Also, no evidence suggests that interference with the apo E-amyloid interaction can produce a clinical benefit for persons predisposed to Alzheimer disease. Nevertheless, the established roles of apo E in plasma lipoprotein metabolism (and now within the central nervous system) make it a serious contender for consideration as one of the biomedical molecules of the year.

1 January 1995 • Annals of Internal Medicine • Volume 122 • Number 1

Robert A. Hegele, MD St. Michael's Hospital Toronto, Ontario M5B 1W8 Canada

Farooq A. Padder, MD Puneet Sahgal, MD Ernesto E. Jonas, MD Nassau County Medical Center East Meadow, NY 11554

References 1. Davidson NO, Carlos RC, Drewek MJ, Parmer TG. Apolipoprotein gene expression in the rat is regulated in a tissue-specific manner by thyroid hormone. J Lipid Res. 1988;29:1511-22. 2. Salter AM, Hayashi R, Al-Seeni M, Brown NF, Bruce J, Sorenson O, et al. Effects of hypothyroidism and high-fat feeding on mRNA concentrations for the low-density-lipoprotein receptor and on acyl-CoA: cholesterol acyltransferase activities in rat liver. Biochem J. 1991;276:82532. 3. Sanan DA, Weisgraber KH, Russell SJ, Mahley RW, Huang D, Saunders A, et al. Apolipoprotein E associates with beta amyloid peptide of Alzheimer's disease to form novel monofibrils: isoform apoE4 associates more efficiently than apoE3. J Clin Invest. 1994;94:860-9. 4. Strittmatter WJ, Saunders AM, Schmechel D, Perciak-Vance M, Enghild J, Salvesen S, et al. Apolipoprotein E: high avidity binding to beta-amyloid and increased frequency of type 4 allele in late-onset familial Alzheimer's disease. Proc Natl Acad Sci U S A . 1993;90:197781. 5. Nathan BP, Bellosta S, Sanan DA, Weisgraber KH, Mahley RW, Pitas RE. Differential effects of apolipoproteins E3 and E4 on neuronal growth in vitro. Science. 1994;264:850-2.

Clinical Use of Apolipoprotein Quantitation To the Editor: In their excellent review on plasma apolipoproteins, Rader and colleagues (1) recommend that fasting determinations of both apolipoprotein B (apo B) and lipoprotein(a) [Lp(a)] levels should be made in persons with coronary artery disease who are not candidates for pharmacologic therapy as determined by low-density lipoprotein (LDL) cholesterol levels less than 3.4 mmol/L (130 mg/dL). Within this group, the patients with elevated levels of apo B, Lp(a), or both are considered candidates for drug therapy. However, in the authors' algorithm, the patients with normal levels of apo B and Lp(a) and low levels of high-density lipoprotein (HDL) cholesterol are still considered candidates for drug therapy. This algorithm should first recommend a determination of HDL cholesterol levels in patients with coronary artery disease and LDL cholesterol levels less than 3.4 mmol/L (130 mg/dL). According to the authors' algorithm, those with low HDL cholesterol levels would be considered candidates for drug therapy, regardless of their levels of apo B or Lp(a). In the patients with normal HDL cholesterol levels, the determination of apo B and Lp(a) levels would then identify those for whom drug therapy would be recommended. The same consideration can be applied to Rader and colleagues' algorithm for persons with a family history of premature coronary artery disease and LDL cholesterol levels between 3.4 and 4.9 mmol/L (130 to 190 mg/dL). Those with low HDL cholesterol levels would be considered for drug therapy regardless of their levels of apo B and Lp(a); determinations of apo B and Lp(a) would only be recommended for persons with normal HDL cholesterol levels. These changes to the authors' algorithms will significantly reduce the cost of diagnostic laboratory tests without sacrificing the quality of the decision-making process. Federico R. Justiniani, MD Mt. Sinai Medical Center Miami Beach, FL 33140 Reference 1. Rader DJ, Hoeg JM, Brewer HB Jr. Quantitation of plasma apolipoproteins in the primary and secondary prevention of coronary artery disease. Ann Intern Med. 1994;120:1012-25. To the Editor: In their excellent review article, Rader and colleagues (1) stated that a 10% decrease in apo B was associated with a 22% increase in coronary artery disease mortality. This result would appear contrary to the expected correlation derived from primary prevention trials such as those of the Lipid Research Clinics (2, 3). How can this finding be explained?

References 1. Rader DJ, Hoeg JM, Brewer HB Jr. Quantitation of plasma apolipoproteins in the primary and secondary prevention of coronary artery disease. Ann Intern Med. 1994;120:1012-25. 2. The Lipid Research Clinics Coronary Primary Prevention Trial results. I. Reduction in incidence of coronary heart disease. JAMA. 1984;251: 351-64. 3. The Lipid Research Clinics Coronary Primary Prevention Trial results. II. The relationship of reduction in incidence of coronary heart disease to cholesterol lowering. JAMA. 1984;251:365-74. To the Editor: The recent review of plasma apolipoproteins by Rader and colleagues (1) was thorough and generally excellent. We feel, however, that several points deserve comment. First, there are now international reference materials for apo A-I and apo B (2, 3). Manufacturers of immunochemical reagents and their related assay kits are currently converting to the values assigned to these reference materials. Admittedly, some differences will persist among methods and laboratories, but these should be substantially reduced after everyone has converted to the new values. Second, the authors state that the British United Provident Association (BUPA) data (4) do not support using apo A-I levels compared with HDL cholesterol levels in population-based screening for coronary heart disease. However, the study showed that these two analytes did not significantly differ in screening performance and that neither was a good screening measurement. Third, the statements about levels of apo B and Lp(a) found in the BUPA study are incorrect. A 10% decrease in apo B levels was associated with a 22% decrease (not increase) in the coronary artery disease mortality rate, and a 10% increase in Lp(a) levels was associated with a 3% increase (not decrease) in the risk for coronary artery disease. One of the major conclusions of the BUPA study was that none of the lipid and lipoprotein variables measured—including levels of total cholesterol, triglycerides, HDL cholesterol, apo A-I, apo B, and Lp(a)—meet reasonable standards for use in population-based screening for coronary heart disease, either alone or in combination. The main conclusion of the study was that population-wide risk reduction is more likely to be effective in reducing the incidence of coronary heart disease than is an attempt to screen for "at-risk" persons (4). A. Myron Johnson, MD The Moses H. Cone Memorial Hospital Greensboro, NC 27401-1020 Wendy Y. Craig, PhD Thomas B. Ledue, BA Foundation for Blood Research Scarborough, ME 04070-0190 References 1. Rader DJ, Hoeg JM, Brewer HB Jr. Quantitation of plasma apolipoproteins in the primary and secondary prevention of coronary artery disease. Ann Intern Med. 1994;120:1012-25. 2. Marcovina SM, Albers JJ, Henderson LO, Hannon WH. International Federation of Clinical Chemistry standardization project for measurements of apolipoproteins A-I and B. III. Comparability of apolipoprotein A-I values by use of international reference material. Clin Chem. 1993;39:773-81. 3. Marcovina SM, Albers JJ, Kennedy H, Mei JV, Henderson LO, Hannon WH. International Federation of Clinical Chemistry standardization project for measurements of apolipoproteins A-I and B. IV. Comparability of apolipoprotein B values by use of international reference material. Clin Chem. 1994;40:586-92. 4. Wald NJ, Law M, Watt HC, Wu T, Bailey A, Johnson AM, et al. Apolipoproteins and ischaemic heart disease: implications for screening. Lancet. 1994;343:75-9. In response: In our proposed algorithms, HDL cholesterol levels are initially determined as part of the fasting lipid panel. We


69

agree that a decision to use drug therapy because of a low HDL cholesterol level could obviate the need for further apolipoprotein quantitation. However, because drug therapy for a low HDL cholesterol level is not established practice, we feel that the additional data provided by the apo B and Lp(a) levels could be useful in determining drug therapy for persons with low HDL cholesterol levels. Two unfortunate misprints appeared in our review. As Dr. Padder and colleagues and Dr. Johnson and colleagues pointed out, a 10% decrease in apo B levels was associated with a 22% decrease in coronary artery disease mortality, and a 10% increase in Lp(a) levels was associated with a 3% increase in the risk for coronary artery disease. Finally, we applaud the international efforts to standardize apo A-I and apo B assays and look forward to when they will be fully implemented. We agree that the BUPA data support neither apo A-I nor HDL cholesterol as a measurement for population-based screening for coronary heart disease but cannot readily explain the discrepancy between the findings of this study and those of others. Daniel J. Rader, MD University of Pennsylvania Medical Center Philadelphia, PA 19104 Respirators and Tuberculosis To the Editor: Nettleman and colleagues (1) have provided an important contribution to the literature on the cost-effectiveness of respirators in preventing transmission of tuberculosis. We are concerned that they present a worst-case scenario and do not assess the use of respirators as part of a hierarchy of industrial hygiene interventions directed at risk reduction. This unfortunately limits the applicability of their results. For example, what if respirators were used for only some health care workers, visits to rooms were limited, reusable negative-pressure cartridge respirators were substituted (cost per use, $0.10 or less), patients were discharged in fewer than 14 days with directions on "source control" of coughing, changes in sputum collection procedures were implemented to allow for more rapid exclusion of uninfected patients from isolation, case identification were improved to limit unnecessary isolation, and purified protein derivative surveillance rates within the Veterans Affairs system were increased to ensure that estimates of employee risk could be adequately assessed? With these considerations in mind, we suspect that a national program to control tuberculosis transmission that places respirator use in the context of other administrative, source, and engineering controls may justify the costs associated with ensuring worker health. In addition, changes in criteria for respirators proposed by the National Institute for Occupational Safety and Health to meet the proposed standard may further decrease the economic burden. We wish to offer a correction to the statement that "no evidence suggests that their use has ever prevented a single case of tuberculosis." Tuberculosis infection results from airborne spread of infectious particles. Respirators are designed to efficiently filter particles the size of droplet nuclei that contain Mycobacterium tuberculosis. Excellent data support the potential for wellfitted respirators to filter out droplet nuclei and to prevent infection. Finally, epidemiologic and experimental studies that document the efficacy of respirators do not exist. They are not likely to be done because of the ethical and logistic (sample size) constraints posed by such studies and because of the multiple reports of high incidence rates of tuberculosis among health care workers in some situations. Perhaps the more important questions to ask about respirator use are where, when, and in conjunction with which interventions should respirators be used to reduce workplace hazards? Scott Barnhart, MD, MPH Nancy Beaudet, MS University of Washington Seattle, WA 98104 Reference 1. Nettleman MD, Fredrickson M, Good NL, Hunter SA. Tuberculosis control strategies: the cost of particulate respirators. Ann Intern Med. 1994;121:37-40.

70

To the Editor: The illuminating article by Nettleman and colleagues (1) places the cost of particulate respirator use in preventing tuberculosis in Veterans Administration hospital workers at $7 million per case prevented and $100 million per life saved. They assumed a case mortality rate of 1% on the basis of the national 1992 incidence and mortality rate figures. That mortality rate, however, is heavily concentrated in patients with the acquired immunodeficiency syndrome, homeless persons, alcoholic persons, and those born outside the United States. The case mortality rate among cooperative and well-nourished Veterans Administration health care workers should be less than 1%. If the case mortality rate among employees treated for tuberculosis is 1%, the cost per life saved by particulate respirators becomes $660 million. A more productive use of this vast sum can surely be devised by public health authorities. Harold L. Israel, MD, MPH Thomas Jefferson University Philadelphia, PA 19107 Reference 1. Nettleman MD, Fredrickson M, Good NL, Hunter SA. Tuberculosis control strategies: the cost of particulate respirators. Ann Intern Med. 1994;121:37-40.

To the Editor: Nettleman and colleagues (1) presented concerns about the high costs of using particulate respirators in preventing transmission of tuberculosis (including multidrug-resistant tuberculosis) to health care workers exposed to patients with tuberculosis in health care facilities. Despite infection control measures, some health care workers have already died from multidrug-resistant tuberculosis. As the number of cases of AIDS and multidrug-resistant tuberculosis increases, we expect to see more deaths among both immunosuppressed and immunocompetent health care workers. In this context, the effectiveness of high-efficiency particulate respirators certainly bears scrutiny. Neither the Centers for Disease Control and Prevention nor the National Institute for Occupational Health and Safety has recommended use of respirators alone; they are part of a comprehensive program that begins with minimizing the number of active cases by doing skin tests and by treating high-risk populations and that includes isolation, negative-pressure rooms, and other administrative and engineering controls in health care facilities. In our facility, I am aware of at least one purified protein derivative test conversion in an immunocompetent third-year medical student who was caring for a poorly compliant patient with multidrug-resistant tuberculosis. The patient was isolated in a negative-pressure room with ultraviolet lights; surgical masks were consistently used for respiratory protection. Our Tuberculosis Task Force is working to provide optimal protection to all employees, with precautions including the use of high-efficiency respirators. Many—perhaps most—health care facilities have few cases of active tuberculosis, and thus the cost of preventive measures will be high in relation to the number of cases prevented. Further research and experience in the use of respirators to prevent tuberculosis may provide methods to save money without compromising protection. The use of particulate respirators currently seems a prudent and practical component to an overall program, despite its considerable expense. William S. Beckett, MD, MPH Yale University School of Medicine New Haven, CT 06510 Reference 1. Nettleman MD, Fredrickson M, Good NL, Hunter SA. Tuberculosis control strategies: the cost of particulate respirators. Ann Intern Med. 1994;121:37-40.

In response: As discussed by Dr. Barnhart and Ms. Beaudet and as stated in our article (1), the number of respirators used could be reduced through reuse or by restricting access to patient rooms. Unfortunately, these measures will have a limited effect. Respirators cannot be reused indefinitely. Most hospitals cannot afford to have a dedicated caretaker serve meals, clean rooms, give medications, respond to patient requests, and provide baths


and examinations. The results of our study, including the number of respirators required per patient-day, were strikingly similar to those reported in a study from the University of Virginia (2). Dr. Israel correctly states that the mortality rate for tuberculosis will vary according to the presence of comorbid conditions. The mortality rate we chose was the figure used for tuberculosis in developing countries and represents an average rate (3). More rapid and effective diagnostic tests would be beneficial but are likely to be expensive and thus not routinely used. We examined the standard methods (three sputum samples stained and cultured for acid-fast bacilli). Dr. Beckett points out that a risk for tuberculosis exists even after patients have been placed in isolation. We discussed the theoretical advantages of the high-efficiency particulate respirator and intentionally used a best-case estimate of its efficacy. Essentially, previous isolation measures (with caretakers using a surgical mask) were assumed to be totally ineffective, and the respirator was assumed to be completely effective. It is unlikely that respirators are this effective. We were correct in stating that respirators have not been shown to prevent a single case of tuberculosis. No study has documented reduced transmission to health care workers through the use of high-efficiency particulate respirators. Why should we choose this moment to abandon our science? If use of particulate respirators is considered unavoidable, why was there not a system established to measure their efficacy? Barnhart and Beaudet argue that the respirators are highly effective (theoretically) and that rates of tuberculosis in employees are "high," but state that prospective efficacy studies are not feasible because of "sample size" considerations. Given the first two assumptions, an efficacy study should be feasible. We owe it to our patients, employees, and visitors to study the efficacy of prevention measures. Perhaps the most important question is how best to use our limited resources to minimize tuberculosis transmission. Mary D. Nettleman, MD, MS University of Iowa College of Medicine Iowa City, IA 52242 References 1. Nettleman MD, Fredrickson M, Good NL, Hunter SA. Tuberculosis control strategies: the cost of particulate respirators. Ann Intern Med. 1994;121:37-40. 2. Adal KA, Anglim AM, Palumbo CL, Titus MG, Coyner BJ, Farr BM. The use of high-efficiency particulate air-filter respirators to protect hospital workers from tuberculosis: a cost-effectiveness analysis. N Engl J Med. 1994;331:169-73. 3. Centers for Disease Control and Prevention. Estimates of future global tuberculosis morbidity and mortality. MMWR Morb Mortal Wkly Rep. 1994;49:961-4.

Cutaneous Alternariosis after Renal Transplantation To the Editor: Alternariosis is an opportunistic fungal disease developing most commonly in immunocompromised patients (1). Although a few cases of deep-seated infections have been described, the disease is usually confined to the skin (2). Proposed treatments consist of systemic administration of amphotericin B or imidazole compounds (3). We report the case of a 50-year-old woman who received a renal allograft in 1991. Immunosuppression was maintained with cyclosporine, azathioprine, and prednisone. One year after the transplantation and 2 months after a local trauma, a purple, crusted, Kaposi-like, painless lesion of the leg developed (Figure 1). Histologic examination showed a granulomatous infiltrate of the upper dermis surrounding septate hyphae that were periodically acid-Schiff-positive in histiocytes. The fungus was identified in culture as related to the genus Alternaria. Serologic results were positive (100 kU/L; normal,