Developing the Tools for Tuberculosis Elimination
At the current rate of decline, approximately 6 percent per year, it will take more than 70 years to reach the target for elimination of tuberculosis of 1 case of tuberculosis per million population. Even if the rate of decline is accelerated to 10 percent per year it will take about 50 years to reach the target. New tools are clearly needed to accelerate the rate of decline to one at which elimination becomes a realistic objective in the next 2 or 3 decades. A great deal of attention has been focused on the development of a new vaccine for the prevention of tuberculosis infection. However, because so many of the cases in the United States are not the result of recent transmission of tuberculosis but rather are the result of reactivation of latent infection, the greatest needs in the United States are new diagnostic tools for the more accurate identification of individuals who are truly infected and who are also at risk of developing tuberculosis. Together with new treatments—drugs or immunological adjuvants—for the prevention of disease in infected individuals, that are easily administered, the elimination of tuberculosis can be a reality. New drugs will also be needed for the treatment of disease both to overcome multidrug resistant strains of tuberculosis and to shorten and simplify current treatment regimens. As all of these new tools will likely contribute to the global fight against tuberculosis, their applicabilities need to be studied. This chapter reviews the current tuberculosis research efforts and outlines strategies for improvement.
Recommendation 5.1 To advance the development of tuberculosis vaccines the committee recommends that the plans outlined in the Blueprint for Tuberculosis Vaccine Development, published by the National Institutes of Health (NIH) in 1998, be fully implemented.
Recommendation 5.2 To advance the development of diagnostic tests and new drugs for both latent infection and active disease, action plans should be developed and implemented. The Centers for Disease Control and Prevention (CDC) should then exploit its expertise in population-based research to evaluate and define the role of promising products.
Recommendation 5.3 To promote better understanding of patient and provider nonadherence with tuberculosis recommendations and guidelines a plan for a behavioral and social science research agenda should be developed and implemented.
Recommendation 5.4 To encourage private-sector product development, the global market for tuberculosis diagnostic tests, drugs, and vaccines should be better characterized and access to these markets for these new products should be facilitated.
Recommendation 5.5 To define the applicability of any new tools to the international arena and facilitate their development, the U.S. Agency for International Development (AID), NIH, and CDC should build upon international relationships and expertise to conduct research.
BACKGROUND AND INTRODUCTION
In 1989, the CDC Advisory Council for the Elimination of Tuberculosis (ACET) called for the development and evaluation of new technologies for tuberculosis diagnosis, treatment, and prevention and the rapid transfer of newly developed technologies into clinical and public health practice. The International Task Force echoed this sentiment for disease eradication and concluded in 1993 (Centers for Disease Control and Prevention, 1993) that tuberculosis could not be eradicated without better tests, treatments, case findings, and vaccines. In the years that have passed since these pronouncements, progress has been made on all of these fronts, but much remains to be done. This chapter reviews the current status and
needs for tuberculosis diagnosis, treatment, and prevention, the dynamics that influence research efforts, and the ongoing activities of the key groups involved. It also identifies the research efforts needed to accelerate the decline in cases and the move toward the elimination of tuberculosis.
CURRENT STATUS AND NEEDS
Currently, the management of tuberculosis is dependent on the ability to identify individuals with active disease, those whose disease is caused by organisms resistant to antimicrobial agents, those who are infected but not ill, and those who are most likely to progress from infection to disease. Although precise numbers are not known, the World Health Organization (WHO) (Mark Perkins, personal communication) estimates that each year sputum examiners screen approximately 60 million to 80 million people for active tuberculosis and that multiple examinations are conducted for each of these people. However, even gross estimates of the number of tuberculin skin tests, for acid-fast bacilli (AFB) smears, radiographs, cultures, or other tests are not available. Although great strides have been made in decreasing the burden of tuberculosis in the United States by using existing technologies, there is considerable room for improvement in each of these endeavors, particularly in children, human immunodeficiency virus (HIV)-infected individuals, and the increasing numbers of patients with tuberculosis outside of their lungs.
The microscopic examination of sputa for the detection of AFB (AFB smear) is rapid, technically simple, and widely available and identifies those thought to be most infectious to others. However, WHO estimates that AFB is identified by sputum microscopy in only 35 percent of people with active tuberculosis (Raviglione et al., 1997). Furthermore, a molecular epidemiological approach has demonstrated that even in efficiently administered tuberculosis control programs, those persons who are AFB smear negative account for at least 15 percent of cases of disease transmission (Behr et al., 1999). Laboratory techniques for the replacement or improvement of the detection of AFB in clinical specimens that address these limitations would greatly enhance patient care and infection control.
Cultivation of mycobacteria from specimens obtained from persons suspected of having tuberculosis remains the mainstay for the diagnosis of tuberculosis and the identification of bacterial strains that are resistant to antibiotics. The use of rapid radiometric techniques has greatly decreased the time required for cultivation and susceptibility testing of mycobacteria, and the use of DNA probes can speed the identification of
organisms once they are growing in culture (Crawford, 1994). However, these techniques still require an average of 10 to 14 days and are generally available only in reference laboratories. The delay in reporting because specimens are being evaluated through a cascade of different laboratories can be long.
Nucleic acid amplification tests for the direct detection of Mycobacterium tuberculosis in clinical specimens remain one of the unfulfilled promises in tuberculosis. Eagerly anticipated nucleic amplification assays that can detect the presence of M. tuberculosis in hours, such as the polymerase chain reaction (PCR), are now licensed. However, the fact that uncertainties remain about the appropriate use of such techniques in clinical practice, despite numerous peer-reviewed publications on this topic, illustrates the challenges in assessing the strengths and limitations of new diagnostic methods and defining their appropriate role in the context of classic diagnostic algorithms and diverse clinical settings (Anonymous, 1997). This vast yet inadequate literature demonstrates the manner in which a lack of coordination between basic scientists and industry can yield an obstructed pipeline of product development. Defining the rational use of existing tests and streamlining the evaluation of new tests are clear and present needs.
Over the past decade, bacterial genotyping techniques have become routine for the identification and tracking of individual strains and have greatly enhanced the ability to detect point source outbreaks and laboratory cross-contamination. However, a demonstrable impact of these on disease control is lacking. If molecular biological methods could be developed for the identification of specific strains that are highly transmissible or that have an enhanced capacity to cause progression to active disease in the host, public health officials could more efficiently focus efforts on contact identification and treatment of latent infection.
In contrast to recent advances in clinical mycobacteriology, there have been no improvements in the ability to identify persons who are latently infected with M. tuberculosis. Tuberculin skin testing with purified protein derivatives from M. tuberculosis has remained essentially unchanged for more than 60 years. This test requires trained personnel for administration and interpretation, two patient encounters within 2 to 7 days, and is falsely negative for approximately 20 percent of individuals with active disease and a greater percentage of the growing numbers of persons coinfected with HIV (Holden et al., 1971; Nunn and Douglas, 1980). Because of cross-reactivity with mycobacteria other than M. tuberculosis and Mycobacterium bovis, complex algorithms of different clinical cutoff points continue to confuse health care providers (Huebner et al., 1993). New reagents or assays are needed to address these limitations (Streeton et al., 1988). Whole genome approaches provide the promise of identifying
species-specific antigens, but exploitation of this approach and validation of new tests will be methodologically and logistically challenging.
Rational targeting of treatment to those individuals infected with M. tuberculosis is hampered by the inability to reliably predict who among them will progress to active disease. Although the clinical and epidemiological characteristics of patients likely to progress to active disease (such as silicosis, diabetes, injection drug use, and low weight) have been discussed for decades, the majority of individuals who do progress to active disease lack these characteristics. Thus, the current proposal is the treatment of all individuals identified to have latent infection, even though active infection will be reactivated in only 10 percent or less of these individuals. Evidence of host genetic factors and surrogate markers of protection would permit targeted interventions.
A new method for the diagnosis of latent infection should be highly specific and ideally will identify those individuals who harbor viable tubercle bacilli or who are at risk for active tuberculosis in the future. In Chapter 4 , selective skin testing by country of origin based on the prevalence of infected individuals in that country is suggested. A more specific test would permit a more broad-based screening as the possibility of false-positive reactions is reduced. However, even among those who are most recently infected and disease-free, only 10 to 15 percent will develop disease in the future. The ability to distinguish that group from those who have developed an immune response that will prevent the development of disease will greatly enhance the efficiency of any effort to prevent disease among latently infected individuals. Even with the regimens currently available for the treatment of latent infection, a highly specific test that detects individuals at the highest risk of developing disease in the future would have a greater impact on the elimination of tuberculosis in the United States than a vaccine that prevents infection. Such a test might also make the treatment of latent infection in many countries with medium and low average incomes a practical intervention.
Once identified, persons who are either ill from M. tuberculosis infection or infected with M. tuberculosis can be treated with a high degree of success. However, current approaches to the treatment of tuberculosis have serious limitations. Treatment of tuberculosis currently requires the administration of multiple different agents for a minimum of 6 months. Success in the treatment of disease in the growing number of persons with disease caused by strains of M. tuberculosis resistant to antibiotics is less certain, requiring a considerably longer duration of therapy and considerable toxicity. The treatment of asymptomatic tuberculosis infection
with potentially toxic drugs is based on an algorithm so complex as to be misunderstood by most health care providers and thus is often applied to less than half of the eligible individuals.
Since streptomycin, the drug first effective in the treatment of tuberculosis, was released and approved for general clinical study and use in 1948, a number of other antimicrobial agents have been approved for use in this disease (See Table 1 in Appendix F). Decades of coordinated clinical trials have demonstrated that these agents can be used in a dizzying number of regimens (Iseman and Sbarbaro, 1991). However, treatment of tuberculosis still requires the simultaneous administration of multiple different drugs for protracted periods. Given the pharmacokinetics of agents approved for use in the United States, most of these drugs require frequent dosing, and cure of tuberculosis now requires that a patient consume a minimum of seven tablets each day.
Unfortunately, these established treatment regimens are increasingly ineffective because of the increasing incidence of tuberculosis due to drug-resistant M. tuberculosis in the United States and around the world. For example, in a survey of M. tuberculosis isolates obtained from 13,511 patients in the United States between 1994 and 1997, 12.3 percent had drug resistance among newly diagnosed and previously untreated patients to any (isoniazid, rifampin, streptomycin, ethambutol) drug, 8.2 percent had primary resistance to one drug, and 4.1 percent had primary resistance to more than one drug (Pablos-Méndez et al., 1998). In the same survey the prevalence of drug resistance among an additional 833 isolates from previously treated individuals in the United States was considerably higher: 23.6 percent were resistant to any drug, 12.5 percent were resistant to one drug (8.1 percent were resistant to isoniazid), 11.2 percent were resistant to more than one drug, and 2.0 percent were resistant to four drugs. Problems of drug resistance are considerably greater in other parts of the world. For example, among isolates from 575 patients in Latvia, 41.6 percent were resistant (combined primary and acquired) to any drug, 34.7 percent were resistant to more than one drug, and 7.0 percent were resistant to all four drugs (Pablos-Méndez et al., 1998). There are limited data on treatment outcomes for subsets of patients with specific single-drug resistance.
Most troubling is the emergence of multidrug-resistant strains that are resistant to at least the two most powerful drugs, isoniazid and rifampin. In contrast to single-drug-resistant strains, which can be effectively treated with prolonged courses of alternative drugs, the success of treatment of multidrug-resistant tuberculosis is only marginally better than outcomes of tuberculosis in the prechemotherapy era. In a report on global surveillance for antituberculosis drug resistance, the data for the United States indicated that the combined rates of primary and acquired
multidrug resistance was 2.0 percent (Pablos-Méndez et al., 1998). In India, a country which accounts for about 30,000 immigrants to the United States each year, multidrug-resistant isolates occurred in 13.3 percent of patients. Closer to the United States, data from Mexico suggest multidrug resistance rates of 6 percent (Centers for Disease Control, 1998). More recent data, however, suggest that the early diagnosis and treatment of tuberculosis have resulted in improved outcomes for patients with tuberculosis in Mexico.
The need for both shorter, simpler, and less toxic regimens and regimens for the treatment of disease caused by multidrug-resistant M. tuberculosis provides ample cause for the development of new agents. However, the development of new antituberculosis drugs by pharmaceutical companies in recent decades has faltered and has fallen behind that of other antimicrobial agents, to a large extent because of the perceived relatively limited market of patients with active tuberculosis in the United States (19,851 cases in 1997). During the 5-year period between 1992 and 1996, a total of 42 new antimicrobial drugs (exclusive of topical agents, additional indications for older drugs, and new preparations of older drugs for different routes of administration) were approved by the U.S. Food and Drug Administration (FDA). Of the 42 new antimicrobial drugs, 17 were antibacterial drugs, 14 were antiviral agents;, 4 were antifungal drugs, 5 were antiparasitic drugs, and only 2 were antituberculosis drugs. Indeed, of the latter two drugs, only one, rifabutin, a drug in the rifamycin group (like rifampin), was a new drug; the other was rifater, a combination of three already long-approved drugs (isoniazid, pyrazinamide, and rifampin). Another drug, levofloxacin (the more active isomer of ofloxacin), active against M. tuberculosis, was approved for use as an antibacterial agent, not for the primary treatment of tuberculosis. Another new rifamycin group drug, rifapentine, was approved by FDA in 1998 for the treatment of pulmonary tuberculosis. Its principal difference from rifampin is in its pharmacokinetics (half-life of 13.2 hours, versus a half-life of 3.4 hours for rifampin), allowing twice-weekly dosing during the 2-month intensive dosing phase of multidrug treatment and once-weekly dosing in the four-month continuation phase of the short-course program of therapy. In sum, in the past 7-year period, no fundamentally new anti-tuberculosis drugs have been approved for the treatment of either tuberculosis infection or multidrug-resistant tuberculosis.
Significant advances have been made in understanding the mechanisms by which M. tuberculosis resists antimicrobial agents, but the mechanisms remain to be defined for many agents. Furthermore, none of these insights have been translated into tangible improvements in clinical practice.
Options for the treatment of tuberculosis infection, although recently
expanded (American Thoracic Society, 2000) remain problematic. Isoniazid, the drug for which the most data are available, requires at least 9 months of daily therapy to be maximally effective. Although recently approved, the combination of rifampin and pyrazinamide for 2 months is based on limited data. There are even fewer scientific data for a 4-month course of rifampin. Prospects for new drugs for the treatment of latent infection are hampered by the lack of techniques for the screening of drug activity against latent bacteria. In combination with increasing rates of drug-resistant tuberculosis, treatment of tuberculosis infection is particularly problematic. It has been inferred, although never rigorously documented, that resistance to isoniazid negates the efficacy of treatment of latent infection with isoniazid. Similarly, newer regimens with rifampin and pyrazinamide are likely to be relatively ineffective against multidrug-resistant strains.
Even when effective regimens are available for the treatment of infections caused by drug-resistant organisms, it is unclear how regimens will be selected for individual patients since it is generally impossible to know the susceptibility of the latent bacilli that they harbor. A lack of understanding of immunity hampers efforts to develop immunomodulators, although thalidomide is a promising proof of principle for the concept of immune modulator in tuberculosis treatment.
The only available vaccine against tuberculosis is BCG, originally developed by Calmette and Guérin by attenuation through serial passage of an M. bovis strain 230 times in vitro between 1908 and 1921. The protective efficacy rates of this vaccine have varied widely when studied over the past half century. Controlled trials and case-control trials have shown protective efficacy rates of the vaccine have a wide range from less than zero percent (i.e., vaccinated individuals were at higher risk) to 75 to 80 percent (Centers for Disease Control and Prevention, 1996). Postulated reasons for this variability include differences in the rate of background infection with environmental mycobacteria, differences in vaccine strains that developed before seed lots were stabilized in a lyophilized form, and other factors related to the patient's environment and socioeconomic status. Two meta-analyses (10 randomized clinical trials and 8 case-control studies in one meta-analysis and 14 clinical trials and 12 case control trials in the second one) have been performed and indicate that the efficacy of the BCG vaccine against meningeal and miliary tuberculosis in children was high (i.e., 80 to 85 percent), but the efficacy of the vaccine against pulmonary tuberculosis differed markedly between the studies of the first meta-analysis and was sufficiently wide as to preclude determina
tion of a summary protective efficacy rate (Rodrigues et al., 1993). In the second meta-analysis the overall protective efficacy of BCG was calculated as 50 percent (Colditz et al., 1994). When present, the efficacy of vaccination with the BCG vaccine has been shown to persist for at least 10 years (Sterne et al., 1998) and may last for more than 20 years (Aronson et al., 1999).
Much has been written that describes the attributes of an ideal tuberculosis vaccine. The list includes safety and the ability to protect against infection and disease, be easily administered, be long-lasting, be inexpensive, be heat stable, not interfere with tuberculin skin testing, and be easily integrated into existing immunization schedules. A less well explored topic is the mechanism by which candidate vaccines can be evaluated for these variables. The difficulty in conducting clinical trials with vaccines against tuberculosis is exemplified by the fact that many of these attributes remain controversial for BCG after more than 70 years of clinical experience. Given the unlikelihood of identifying a perfect tuberculosis vaccine in the proximate future, it is important to define what would comprise a minimally acceptable vaccine or combination of vaccines to meet global needs.
The needs for and approach to tuberculosis vaccine development were recently elucidated at a workshop convened by the U.S. Department of Health and Human Services, the U.S. National Vaccine Program Office, and NIH (National Institutes of Health, 1998). This workshop emphasized the need for a coordinated, multifaceted, minimal, long-term commitment of $800 million over the coming 20 years. Initial steps proposed a focus on basic research, animal models, clinical trial design, and international capacity building. These will need to be linked to vaccine production facilities and epidemiologically well-characterized U.S. and international field sites.
The WHO Global Tuberculosis Research Initiative has identified social science research as a priority. Earlier a call for a similar agenda had been made in the United States (Centers for Disease Control and Prevention, 1995).
Despite the renewed focus on tuberculosis in the United States, many health care providers remain ignorant about tuberculosis, many health care systems do not prioritize it, and many patients do not adhere to the treatment. Thus, there is a need to understand the determinants of behavior of health care providers and systems (e.g., health maintenance organizations) as well as the behavior of patients and to improve methods for predicting and monitoring patient adherence and compliance with
therapy, particularly in marginalized populations and immigrants. A panel of effective methods that can be used to enhance patient and provider adherence and compliance must be developed and implemented in a targeted fashion. The need to understand the determinants of behavior is particularly acute for the development of programs for the treatment of latent infection. Asking people to take medication for a disease that they do not yet have is likely to present a totally new set of challenges. In addition, the call for programs of treatment for latent infection among foreign-born individuals presents yet another dimension of the need to understand human behavior, as ethnic variations in knowledge, attitudes, and behavior need to be elucidated and understood. There has been a recent effort to understand the costs and economics of tuberculosis, especially through the Prevention Effectiveness Studies Section, in the Research Branch of the Division of Tuberculosis Elimination. This research needs to be expanded and will be central to selecting the most cost-effective strategies as incidence declines. Another area for operational research, contact investigations and outbreaks, is described in Chapter 4 .
DYNAMICS OF TUBERCULOSIS RESEARCH
An evaluation of current and future research activities must recognize and account for the different goals, expertise, priority-setting mechanisms, and evaluation criteria of the diverse partners in tuberculosis research. Such an understanding both will explain current activities and is crucial in any attempt to influence the focus of these efforts. The ultimate goal of implementing an efficient pipeline of research from the “bench to the bedside” can be met by recognizing and influencing these dynamics. These partners can be broadly classified as basic, applied, and operational researchers.
The goal of basic investigators is to advance fundamental knowledge in the biology of tuberculosis. This goal is pursued by highly specialized individuals who are largely working in academic and research institutes using a broad array of modern techniques that have been developed in diverse scientific fields such as bacteriology and immunology. In general, the focus of their efforts is set by individual investigators' perceptions of the importance and opportunities in their fields. The actual research portfolio is heavily influenced by the availability of research funds, as demonstrated by the dramatic expansion of basic tuberculosis research efforts stimulated by recent increases in funding for such research. These investigations are largely evaluated by peer review of grant applications and the publications that result from the research.
Applied researchers aim to translate basic knowledge into pragmatic applications. These efforts can be further dichotomized into early “proof-
of-principle” efforts (largely conducted by investigators in academia and the biotechnology industry) and developmental efforts (almost exclusively conducted by workers in the pharmaceutical and diagnostics industries). As such, although these individuals' priorities are influenced by the availability of public research funds, they are primarily influenced by industrial decision makers on the basis of the perceived economics of the potential market. Their productivity can be evaluated in terms of the patents and products that are ultimately marketed.
The goal of operational researchers is to optimize the efficiency of application of licensed technologies. Thus, operational researchers require an understanding of the realities of the situations in which tuberculosis is being diagnosed (such as the factors that influence health care providers and clients) and an ability to work effectively in partnership with these communities. The major contributors in these efforts are the CDC, WHO, and a small number of academic investigators and members of nongovernmental organizations. Priorities in this realm are set by the potential of their efforts to affect outcomes. As such, they are evaluated by the policies that they establish and ultimately by the effects that these policies have on clinical outcomes.
National Institutes of Health (NIH)
In 1989, a year in which more than 2 million people died from tuberculosis, only a few NIH-funded laboratories were investigating M. tuberculosis. Largely in response to the resurgence of the disease in the United States, the early 1990s was a time of unprecedented expansion in NIH-sponsored tuberculosis research (Figure 5-1).
By 1995, the total NIH budget for tuberculosis research was $62 million. This expansion has been followed by a period of stagnation, in which funding for tuberculosis research increased only slightly faster than the overall rate of inflation. The vast majority of this research funding is administered by the National Institute of Allergy and Infectious Diseases ($41.7 million), and the National Heart, Lung, and Blood Institute ($16.1 million), but also includes the National Institute of Drug Abuse ($6.6 million), the National Center for Research Resources ($4.8 million), Fogarty International Center ($2.2 million), National Institute of Diabetes and Digestive and Kidney Diseases ($1.5 million), National Institute of Child Health and Human Development ($0.8 million), National Institute on Alcohol Abuse and Alcoholism ($0.5 million), National Institute of Mental Health ($0.4 million), and the National Institute of Nursing Research ($0.4 million).
The research sponsored by NIH can be classified into U.S. Public Health Service categories of pathogenesis, diagnostics, therapeutics, vaccines, epidemiology, and training. Data on funding for tuberculosis research from the National Institute of Allergy and Infectious Diseases are presented in Figure 5-2 .
Although most of this work is extramural, with the $35 million award to the Tuberculosis Research Unit being the largest single extramural award, three intramural research programs focus on tuberculosis. The majority of the extramural research is funded through investigator-initiated research grants. In addition, non-hypothesis-driven efforts are funded through contracts focused on three areas: research materials and vaccine testing, a coordinated research unit, and a facility for drug screening.
Training of young scientists in the field of tuberculosis research is a stated priority of NIH and occurs both through programs targeted toward tuberculosis (e.g., the Tuberculosis International Research Training Program) and through generally focused graduate and postdoctoral fellowships. In addition, the National Heart, Lung, and Blood Institute has a specific program targeted toward curriculum development to teach young scientists about tuberculosis.
Although no single parameter accurately assesses the productivity of research, in the realm of basic science, the numbers of papers published
and the number of citations that they receive in other papers or reports is perhaps the best. Figure 5-3 shows the numbers of Medline publications under the heading of “tuberculosis” by year between 1989 and 1998. These publications have been divided according to the 19 relevant MESH subheadings, which have been grouped into five categories (pathogenesis/ vaccines, diagnostics, therapy/complications, epidemiology/control, and other). Of note, the total number of annual publications, which was roughly stable from 1966 to 1989, increased through the early 1990s as funding and public attention was increasing. Importantly, the types of publications have remained balanced. Thus, from this admittedly imperfect measure, one senses that productivity in tuberculosis research has increased and balance has been maintained in the research portfolio.
Centers for Disease Control and Prevention
The research programs of CDC currently encompass a broad range of activities including basic bacteriological research, surveillance, operational research, behavioral studies, technology assessment, and transfer. Although it is difficult to precisely account for funds spent on these efforts (as opposed to related activities), the total research budget is $15 million per year, approximately 10 percent of the overall CDC budget for tuberculosis.
Among this diverse group of activities, several seem to be particu
larly productive. CDC surveillance activities (which incorporate novel technologies such as bacterial genotyping) are among the best in the world. Detailed descriptive analysis of disease risk and burden are available and provide a wealth of material for rational improvements in disease control. The Tuberculosis Trials Consortium (TBTC) initiated in 1995 was instrumental in the licensure of rifapentine in 1998. CDC-sponsored clinical trials provided crucial data on which the new recommendations of use of the combination of rifampin and pyrazinamide for the treatment of tuberculous infection were based. CDC has published approximately 20 peer-reviewed papers each year, roughly equally divided among the areas of basic science, diagnostics, and epidemiology.
One apparent impediment to realizing the potential of CDC research is administrative division between epidemiological and laboratory activities. Epidemiological research and control activities are situated in the National Center for HIV, STD, and Tuberculosis Prevention, in the Division of Tuberculosis Elimination, whereas laboratory activities and research are housed in the National Center for Infectious Diseases, in the Division of AIDS, STD, and Tuberculosis Laboratory Research. Although most CDC staff report that communications are good and improving, the perception of many researchers outside of CDC is that there is often a lack of communication and coordination.
New Tuberculosis Diagnostic Developments by Industry
The most systematic assessment of members of industry with an interest in the development of tuberculosis diagnostic tools has been conducted by Mark Perkins, WHO, director of the Tuberculosis Diagnostics Initiative of WHO (personal communication, 1999). They were identified at international meetings, via word of mouth, in response to spontaneous requests for information or referral of such requests from other agencies, through publication of information on the tuberculosis diagnostic initiative, and in discussions with pharmacists and private physicians in developing countries. As of 1999, 58 commercial agencies or academic researchers with industry developmental support have been identified through this survey. Many of the test developers are small diagnostic companies, some of whom had interacted frequently with WHO regarding specimen acquisition, market understanding, and the design of trials for the assessment of diagnostic tools. The respondents had various levels of willingness to share information, spanning a spectrum, from those willing to speak openly about their development plans to others unwilling to discuss even the format of the tools in development or the target matrix. There are many fewer diagnostic approaches than there are industry or academic test developers. This represents both the use by multiple com
panies of tests with the same type and format (with serological assays that target immunoglobulin directed against a single M. tuberculosis antigen being the most common) and the simple repackaging of single products by more than one company. Companies may market diagnostic tools developed in their own facilities (or under contract) as well as those developed by other companies and licensed for marketing under a different name. The reticence of some companies to disclose the nature and development or production site of their products has made it difficult to determine the extent to which the number of companies identified in the survey to be involved with tuberculosis diagnostic tools is inflated by marketing of duplicate products.
Given the limitations in information and the confidentiality agreements under which much of the information was obtained, it is not possible to list the companies identified in the survey of tuberculosis diagnostics, the specifications of the tests that companies are developing, or in some cases even the type of assay being developed. However, Table 5-1 summarizes the range of tests that were found to be under development.
One proxy for potential industrial interest in tuberculosis is the number of U.S. patents filed. Although this clearly reflects activities funded from a variety of sources and conducted by different types of people, it is an objective measure. Figure 5-4 shows the number of tuberculosis-related U.S. patents issued by year as determined by searching the International Business Machines Corp. Intellectual Property Network (http://patent.womplex.ibm.com/). By searching all fields on the U.S. patents front pages, it is possible to observe a dramatic increase in the number of tuberculosis-related patents issued over time.
Please refer to the page image for an unflawed representation of this content.
New Drug Development by Industry
To determine the current activity in the biotechnology and pharmaceutical industries in the field of new drug discovery, the committee has explored two questions: (1) Has industry reported new lead compounds or drugs in the antimicrobial literature? (2) Are biotechnology and pharmaceutical companies pursuing new drugs that are still in initial phases of development? To answer the first question the committee searched by title for articles that dealt with new antimicrobial agents and that were published in three 2-year periods (1988–1989, 1992–1993, and 1997–1998) in Antimicrobial Agents and Chemotherapy, the leading U.S. journal that deals with new anti-infective drugs. During the aforementioned periods, a total of 374 new (not previously reported) drugs were noted in the titles of 374 articles. Forty percent described antibacterial drugs, 20 percent described antiviral agents, but only 3 percent described antimycobacterial agents. When drugs with primary activity against only Mycobacterium leprae or Mycobacterium avium complex are eliminated, only eight drugs
remain, of which two are rifabutin and ofloxacin, and the remainder represent only 1.7 percent of the total.
To answer the second question 22 major pharmaceutical companies and 7 biotechnology companies (see Table 2 and Table 3 in Appendix F) were contacted by phone and asked if they had any new drugs in the pipeline that might be forthcoming in the next 2, 5, or 8 years for use in the treatment of tuberculosis. The results are quite discouraging regarding current discovery efforts. Only four companies indicated that they had any drugs in the pipeline for the time frame as mentioned above. Only one company noted that it had oxazolidinone compounds undergoing preliminary laboratory investigation and, if results were encouraging, would require 5 to 8 years before they might become available for clinical use (Cynamon et al., 1999). Another company noted that it had oxazolidinone compounds undergoing preliminary laboratory investigation, but it had turned over its rights to develop the drug to the company mentioned in the previous sentence. A third company noted that it has no track record with the development of antituberculosis drugs but had found a group of compounds in an antibacterial screening process that it is sending elsewhere for in vitro testing. The marketing team at this company is unenthusiastic about pursuing such a drug(s) because of the limited U.S. market. Finally, one major pharmaceutical company was recently involved in preliminary work with animals on the use of interleukin 12 (IL-12) as an interferon inducer on the basis of the known actions of IL-12. However, the work was recently abandoned when it was found in an experimental animal model of M. tuberculosis infection that treatment with this biological response modifier paradoxically increased the number of M. tuberculosis bacilli in the lungs.
Among the seven biotechnology companies that the committee contacted, one is pursuing development of a nitroimidizopyrene, which inhibits glycolipid cell wall synthesis (at a step different than that at which it is inhibited by ethambutol). It appears to be bactericidal and active against both replicating and nongrowing organisms. If all goes well with future testing, the company believes that the drug might be available for clinical use in 5 to 8 years. The same company has also been developing a rifamycin analog (rifalizyl) with a longer half-life, but the company is giving up development of this analog because M. tuberculosis strains resistant to rifampin show cross-resistance to it. Another biotechnology company is working on three candidate ethambutol analogs that should be ready for clinical testing in the next 2 to 5 years. Another candidate drug, one that inhibits the shikemic acid pathway, is in the preliminary testing phase and is 5 to 8 years away from possible clinical use. Finally, one company is studying the use of a combination of IL-12 plus three conventional antituberculosis drugs for the treatment of M. tuberculosis or M.
avium-M. intracellulare infection in patients with AIDS. The same company is involved in a current clinical study of the combination of IL-12 plus three conventional antituberculosis drugs for the treatment of tuberculosis in The Gambia. Neither study, if the results prove positive, is expected to lead to clinical use of the drugs before 5 to 8 years hence. It should be noted that this approach involves the auxiliary use of a biological response modifier rather than a new antimicrobial agent.
The cost of drug development and the lack of a market for antituberculosis drugs are often cited as barriers to the development of new drugs for the treatment of tuberculosis. The commonly cited cost for the development of a new drug is $350 million. Even if this estimate is accurate it amortizes the cost of all failed candidate drugs into the cost of the successful drug. This is not a legitimate estimate for an antituberculosis drug, given the amount of public funding that supports drug research. This support not only includes the type of basic research commonly associated with NIH but also includes programs to provide rapid in vitro screens for assessment of the drug activities for candidate compounds, evaluation of attractive candidate drugs with animal models, and most recently, a program to assist in translational research that leads to product development. Domestic and international sites also have clinical trial capabilities that can assist with studies, up through phase III trials. With this kind of assistance, the true cost of developing a drug is probably in the range of $15 million to $30 million (C. Nacy, Sequella Global Tuberculosis Foundation, unpublished data). To address the concern about the market for antituberculosis drugs, detailed studies are needed to characterize the markets outside of the established market economies. One estimate puts the global expenditure for the four main antituberculosis drugs at $800 million to $900 million per year, an amount that would definitely justify the development of new drugs in market terms.
Vaccine Development by Industry
To attempt to answer the question as to whether industry is actively pursuing development of candidate vaccines for tuberculosis prevention, the same 22 major pharmaceutical companies and 7 biotechnology companies that were mentioned above and that were questioned about new drug discovery were queried about any prospective vaccines. Twenty-one of the 22 companies contacted responded. There is clearly a paucity of activity in this area among established drug companies. Only two companies are involved in vaccine development. One is working on both a DNA-based vaccine and a vaccine with a live attenuated strain of the microorganism. In addition to a preventive vaccine, they are also exploring therapeutic use of their candidate vaccine in patients with established
M. tuberculosis infections. The second company is involved in vaccine development only in a very preliminary, exploratory fashion. It is working with a DNA-based vaccine that is in the public domain and is also interested in evaluating possible subunit vaccines.
Two of the seven biotechnology companies contacted indicated that their major activities are focused on the prevention or treatment of tuberculosis. For one company, the approach is exclusively vaccine development. It has developed a vaccine that uses three to five protective protein antigens made through the incorporation of the genes for these antigens (effectively, a synthetic gene) in one recombinant protein. The efficacy of this vaccine has been shown in a rodent model, and the safety of the vaccine has been demonstrated with monkeys. It expects clinical trials (phases I and II) to begin in about a year in several countries where the rate of endemic tuberculosis is high. The second biotechnology company is involved in the development of chemotherapeutic agents (ethambutol analogs and shikemic acid pathway inhibitors) as well as a candidate vaccine for the prevention of tuberculosis. This company is working with the experimental DNA vaccine described earlier (Lowrie et al., 1999).
In general, the time frame for full development, clinical testing, and final product marketing (if a successful vaccine were to be forthcoming) would be in the order of 10 to 20 years.
Although the funding for research provided by private foundations is significantly less than that provided by government and industry, philanthropic foundations fill a crucial catalytic niche in many realms of medical research. Total philanthropic giving for 1997 came to $143.46 billion, with $121.89 billion from individuals and bequests, $13.37 billion from noncorporate foundations, and $8.2 billion from corporate foundations. Only 8.8 percent (just over $14 billion) of the total contributions received from charities was in the health category. These contributions included support for health services and health facilities, such as hospitals and nursing homes; support for organizations that address general health or specific diseases; and a small portion of support for medical research.
Grant-making organizations contributed $19.46 billion to nonprofit organizations in 1998, representing a 22 percent increase over 1977 contributions. Foundations currently spend 16.2 percent of total grant dollars, or $1.2 billion, on health. The majority of this (58 percent) goes to general and rehabilitative services, including hospitals and medical care, reproductive health, public health, health policy, and management. Only about $265 million, or 22 percent of the portion that foundations spent on health in 1998, went to medical research. In contrast, in 1998, government sup
port for medical research was about $15 billion and industry support was $18 billion.
Although the amounts are relatively small, these funds provide crucial venture capital essential to the progress of medical research in many fields. In particular, private funders can move quickly to address new needs (particularly in translational research), take risks that potentially have high payoffs (for example, in behavioral research), and attempt novel funding approaches that may serve as a model for other funders.
Although comprehensive data that permit an examination of the categories of medical research that are funded are not available, an informal evaluation by the committee indicated that targeted support for tuberculosis is quite limited in light of the scope of the problem (Bond et al., 1998; Scott, 1999). The recent announcement by the Bill and Melinda Gates Foundation of grants for $50 million over 5 years for tuberculosis vaccine development may signal the start of interest in tuberculosis research by philanthropic foundations.
United States Agency for International Development
In light of the stated humanitarian and development goals of AID and its capacity to translate research advances into practice, its activities in tuberculosis have been extremely disappointing. Reducing the burden of disease would be of clear humanitarian benefit and would foster the sustained prosperity of nations with a high incidence of tuberculosis. However, AID has no tuberculosis program, and the past funding for tuberculosis that has been available has come through the HIV infection and AIDS program. One major AID activity has been the financial support of the global Stop TB Initiative. These funds have begun to build a framework for cooperation and collaboration among countries with a high incidence of tuberculosis, international organizations, and donors. A New Infectious Disease Initiative within AID has identified tuberculosis as one of its four components. However, a rational plan for the prioritization and implementation of these activities remains to be elucidated.
Food and Drug Administration
FDA has both intramural and extramural programs for tuberculosis research, although both are relatively small. The intramural program is centered within the Laboratory of Mycobacteria of the Center for Biologics Evaluation and Research. Funding for this program has steadily declined from about $235,000 in fiscal year 1995 to about $135,000 in fiscal year 2000. However, the program has been maintained through interagency
agreements with NIH and the National Vaccine Program. The laboratory has worked on the generation of plasmids that carry key mycobacterial antigens, early development of vaccines that can be produced by edible plants, and evaluation of the protection conferred by auxotrophic live tuberculosis vaccines and DNA vaccines. Additional work has been conducted on virulence factors and mechanisms of protective immunity. This program, which is staffed by 10 individuals (FDA staff, fellows, and visiting scientists), has generated 24 publications over the last 3 years.
The extramural program is funded through the Orphan Drugs Program. Grants from this program are usually relatively small and are used to support early clinical work on new drugs. The program does not categorically fund tuberculosis research, and any application would be evaluated by external reviewers and ranked against all other applications. The program is not well known, and applications are not actively solicited.
In June 1985, CDC, NIH, the American Thoracic Society, and the Pittsfield Antituberculosis Association co-sponsored a conference to identify priority areas for research that might ultimately lead to the elimination of tuberculosis from the United States. This plan was incorporated into the ACET Strategic Plan for the Elimination of Tuberculosis in the United States (Centers for Disease Control, 1989) and was further elaborated in the Action Plan to Combat Multidrug-Resistant Tuberculosis (Centers for Disease Control and Prevention, 1992) The plan included an extensive list of investigative directions that should be pursued and a plea for their financial support. Reflecting on this list 15 years later is both discouraging and encouraging. It is discouraging to realize that virtually no funding and consequently no research was conducted for 7 years after its publication. It is encouraging to see that, following the expansion of funding that occurred between 1992 and 1995, there has been an explosion of knowledge and the development of a cohort of investigators and facilities in virtually every field of research relevant to tuberculosis control. Yet, the committee concludes that the level of support remains woefully inadequate, considering the global burden of disease and the potential of the current research network to make discoveries that will change the way the prevention, diagnosis, and treatment of tuberculosis are thought of.
The cost for the research plan in the Action Plan to Combat Multidrug-Resistant Tuberculosis was placed at about $160 million per year in 1984. At best the federal funding for research has been about half that amount. Since the development of the action plan, a Blueprint for Tuberculosis Vaccine Development was published 1998. That blueprint proposed a 20-year
Page 144$800 million budget, or about $40 million per year, for vaccine development. The estimate, however, seems to be based on the amount spent annually on the treatment of tuberculosis rather than on an objective costing of the plan. As discussed earlier, the committee recommends a similar initiative for the development of tests for the diagnosis of latent infection that would identify with a high sensitivity and high specificity, those who will develop tuberculosis and another initiative for the development of treatments for those with latent infections (i.e., drugs and immune modulators). That would yield a total budget of about $280 million per year for research to support the original items in the action plan: the vaccine initiative, the diagnostic test initiative, and the treatment initiative. This is the best estimate that the committee can make given the available information. Accurate budget estimates will be needed for all three initiatives to effectively argue for the funding of these programs.
In addition to obtaining appropriate amounts of funding, there are important issues related to the efficient use of those funds. The differences between the different funders and researchers involved in tuberculosis research provide both challenges and opportunities. It will be a challenge to ensure coordination and communication links along the entire pipeline from basic to operational research to maximize efficiency and minimize delays. These differences, however, also provide an unprecedented opportunity to achieve the goals outlined by this committee and prior bodies. Impediments must be replaced with incentives at each step in the process of the development of diagnostic tests, drugs, and vaccines.
Although the committee endorses the scope of the research agenda outlined in the Strategic Plan for the Elimination of Tuberculosis (Centers for Disease Control, 1989) and the clear prioritization relevant to vaccine production delineated in the Blueprint for Tuberculosis Vaccine Development (National Institutes of Health, 1998) (that vaccine development remains an important global priority), the single highest research priority is research relevant to the identification and treatment of tuberculosis infection. This focus is justified in light of the central importance of the diagnosis and treatment of tuberculosis infection to the elimination of tuberculosis from the United States. Although existing evidence and tools justify the expansion of targeted screening and treatment of latent tuberculosis infections now, better tools for the diagnosis of infection and drugs for the treatment of infected individuals would facilitate these activities. The diagnostic test might be approached as a two-step test, with the first step very sensitive and the second step highly specific. An action plan that delineates the short- and long-term strategies for achieving these goals, needs to be fully elucidated by the research community. It is anticipated that this strategy will include the following elements:
The strategy will include NIH because the benefits of basic research cannot be anticipated and are difficult to predict and, thus, it is difficult to make specific recommendations. Nevertheless, the goal of eliminating tuberculosis infection from the United States must be made clear to independent investigators. The best way to ensure that the best minds of the current generation are focused on this goal is to make a sustained commitment to a staged, coordinated tripling of financial support for basic tuberculosis research that is likely to foster this goal. Realms to be supported by these funds must include definition of the bacteriology of “latent” infection and the identification of the pharmacological agents that kill latent bacilli. Specific immunological markers of infection need to be identified and immunological predictors of those individuals with latent tuberculosis who will later have active disease must be found. Funding for mathematical models that can be used to compare various interventions should be encouraged.
Seed grant support for translating basic scientific knowledge into promising tests and compounds must be made available from governmental groups (e.g., through agencies of the Department of Health and Human Services, funding mechanisms such as Small Business Grants, and the Defense Advanced Research Projects Agency). The specifics of the potential market for a novel diagnostic test that can determine who among the millions of Americans at high risk of infection and an effective drug for the treatment of the millions who actually harbor M. tuberculosis must be emphasized to those who set industrial research priorities. Awareness of this market should drive small biotechnology companies into diagnostic assay development and the identification of new agents. The opportunity to test these diagnostic assays and drugs in cooperation with publicly funded networks of clinical trials should entice pharmaceutical and biotechnology companies to expand on the knowledge gained by publicly funded basic investigators and rapidly move new products through commercial development. Formation of a joint venture between industry and public sector agencies, would speed the attainment of this goal. Finally, a detailed study of the global market for licensed tuberculosis pharmaceutical and diagnostic products, along with information on the requirements for entering the largest markets could dispel the conventional wisdom that there is no market for tuberculosis drugs.
A critical reappraisal of the CDC research agenda should be conducted. Its efforts should also be focused on its unique capacity to conduct pragmatic research linking expertise in study design with broad access to patients and clinical specimens to determine the performance characteristics of new laboratory technologies and the efficacies of new drugs. A diagnostics network, akin to the existing Tuberculosis Trials
Consortium, should be developed for the rapid assessment of new tools for the diagnosis of infection and disease. The Tuberculosis Trials Consortium, capable of enrolling large numbers of people each year, should be maintained and expanded to fill the gap between industry-driven product development and product implementation. In the short term, these efforts should be focused on evaluations of currently available agents, such as the use of fluoroquinolones and rifapentine, for treatment of infection. In addition to providing important information, these efforts will prime the pump for new agents as they become available, perhaps oxazolidinones. Intensive monitoring for adverse outcomes as they arise under operational conditions must be pursued before expansion of these programs. Subsequently, this network can be used to provide clinical confirmation of basic insights and advances. This network must reach out to community-based health care providers and for-profit health care providers.
These clinical trials should be extended by AID to include sites in other countries with high rates of tuberculosis infection. These efforts not only would provide important information of relevance to the United States but would also demonstrate the potential of these technologies to improve the health of people in other countries, particularly those that account for considerable proportions of U.S. immigrants.
Operational research must define efficient and cost-effective approaches to the implementation of these approaches, particularly in privatized health care delivery settings and medically marginalized populations.
Another area for increased attention is research related to the epidemiology of tuberculosis as the incidence declines and the disease approaches elimination. The focus of this research should be on distinguishing reactivation of latent infection from more recent transmission and will likely include more widespread genotyping and more traditional types of epidemiological studies. The results of that research should also then be translated into information that will be useful in guiding expanded contact investigations and the identification of outbreak situations.
Elimination of tuberculosis infection is an ambitious goal, but it is central to the elimination of tuberculosis. A sustained commitment to this goal will have considerable incidental benefits.
American Thoracic Society and Centers for Disease Control and Prevention . 2000 . Targeted tuberculosis testing and treatment of latent infection . Am J Respir Crit Care Med 161 : 5221–5247 .
Anonymous . 1997 . Rapid diagnostic tests for tuberculosis . What is the appropriate use? Am J Respir Crit Care Med 155 : 1804–1814 .
Aronson N , Santosham M , Howard R , Comstock G , and Harrison, L. 1999 . The long-term efficacy of BCG vaccine . Abstracts of the 39th Interscience Conference on Antimicrobial Agents and Chemotherapy, September 1999 . Washington, DC : .
Behr MA , Warren SA , Salamon H , Hopewell PC , Ponce de Leon A , Daley CL , and Small PM. 1999 . Transmission of Mycobacterium tuberculosis from patients smear-negative for acid-fast bacilli . Lancet 353 : 444–449 .
Bond EC , Peck MG , and Scott, MB. 1998 . The future of philanthropic support for research . Scientist 13(21) : 15 .
Centers for Disease Control . 1989 . A strategic plan for the elimination of tuberculosis in the United States . MMWR 38 (S3) : 1–25 .
Centers for Disease Control and Prevention . 1992 . National action plan to combat multidrug-resistant tuberculosis in the United States . MMWR 41(RR11) : 1–48 .
Centers for Disease Control and Prevention . 1993 . Recommendations of the International Task Force for Disease Eradication . MMWR 42(RR16) : 1–25 .
Centers for Disease Control and Prevention . 1995 . Improving tuberculosis treatment and control: An agenda behavioral, social, and health services research . Proceedings for “Tuberculosis and Behavior, National Workshop on Research for the 21st Century, 1995.” Atlanta : Centers for Disease Control and Prevention .
Centers for Disease Control and Prevention . 1996 . The role of BCG vaccine in the prevention and control of tuberculosis in the United States . MMWR 45 (RR4) : 1–18 .
Centers for Disease Control and Prevention . 1998 . Population-Based Survey for Drug Resistance of Tuberculosis—Mexico, 1997 . MMWR 47 : 371–375 .
Centers for Disease Control and Prevention . 2000 . Targeted testing and treatment of latent tuberculosis infection . MMWR 49(RR-6) : 1–54 .
Colditz GA , Brewer TF , Berkey CS , et al. 1994 . Efficacy of BCG vaccine in the prevention of tuberculosis . Meta-analysis of the published literature . JAMA 271(9) : 698–702 .
Crawford JT. 1994 . New technologies in the diagnosis of tuberculosis . Semin Respir Infect 9 : 62(70 .
Cynamon MH , Klemens SP , Sharpe CA , and Chase S. 1999 . Activities of several novel oxazolidino nesagainst Mycobacterium tuberculosis in a murine model . Antimicrob Agents Chemother 43 : 1189(1190 .
Holden, M , Dubin MR , and Diamond PH. 1971 . Frequency of negative intermediate-strength tuberculin sensitivity in persons with active tuberculosis . N Engl J Med 285 : 1506–1509 .
Huebner RE , Schein MF , and Bass JB, Jr. 1993 . The tuberculin skin test . Clin Infect Dis 17 : 968–975 .
Iseman MD and Sbarbaro JA. 1991 . Short-course chemotherapy of tuberculosis . Hail Britannia (and friends!) . Am Rev Respir Dis 143(4) : 697–698 .
Lowrie DB , Tascon RE , Bonato VL , et al. 1999 . Therapy of tuberculosis in mice by DNA vaccination . Nature 400(6741) : 269–271 .
National Institutes of Health . 1998 . Blueprint for Vaccine Development . Report of a workshop held March 5–6, 1998, Rockville, MD.
NIAID . 2000 . Blueprint of tuberculosis vaccine development . Clinical Infectious Diseases Vol.30, Supp. 3 (June 2000) , 5233–5242 .
Nunn DR , and Douglas JE. 1980 . Anergy in active pulmonary tuberculosis: A comparison between positive and negative reactors and evaluation of 5 TU and 250 TU skin test doses . Chest 77 : 32–37 .
Pablos-Méndez A , Raviglione MC , Laszlo A , et al. 1998 . Global surveillance for antituberculosis drug resistance, 1994–1997 . World Health Organization-International Union Against Tuberculosis and Lung Disease. Working Group on Antituberculosis Drug Resistance Surveillance . N Engl J Med 338(23) : 164–169 .
Raviglione M , Dye C , Schmidt S , and Kochi A. 1997 . Assessment of worldwide tuberculosis control . Lancet 350 : 624–629 .
Rodrigues LC , Diwan VK , and Wheeler JG. 1993 . Protective effect of BCG against tuberculosis meningitis and miliary tuberculosis: A meta-analysis . Int J Epidemiol 22(6) : 1154–1158 .
Scott, MB. 1999 . The Scientist Oct 25 : 13 .
Sterne JAC , Rodrigues LC , and Guedes IN. 1998 . Does the efficacy of BCG decline with time since vaccination? Int J Tuberc Lung Dis 2 : 200–207 .
Streeton JA , Desem N , and Jones SL. 1998 . Sensitivity and specificity of a gamma interferon blood test for tuberculosis infection . Int J Tuberc Lung Dis 2 : 443–450 .