. THE MEEHHINJZHTJDIN HUTDI‘IIHTJDIN ‘ BIND JINEHEHSED EFFEETJUEINESS 3 BF THE % HSUMJCHL QBEUHHTDHV OF THE EEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE THE MECHANIZATION, AUTOMATION, AND INCREASED EFFECTIVENESS OF THE CLINICAL LABORATORY A STATUS REPORT BY THE AUTOMATION IN THE MEDICAL LABORATORY SCIENCES-REVIEW COMMITTEE of the NATIONAL INSTITUTE OF GENERAL MEDICAL SCIENCES NATIONAL INSTITUTES OF HEALTH 1971 DHEW Publication No. (NIH) 72-145 1.35; 1-: HEALTH IJSRAEY 76837 TABLE OF CONTENTS U~ 4‘8 :2; PAGE Foreword ................................................. .E?%fi.. 3 Contributors ................ . .................... . ............... 4 The Clinical Laboratory ................................. . ........ 5 Instrument Requirements ...................................... 7 Sample Identification ....................................... 10 Data Reduction .............................................. 13 Computer Emulation ........................... . .............. l6 Miniaturization ............................................. 18 Cost effectiveness .......................................... 22 ‘ Clinical Chemistry ............................................... 24 Methodology‘r ................................................. 24 Methods for ”automating” established analytical procedures... 26 Development of improved separation or discrimination techniques, and new identification or estimation techniques. 31 Special problems ...................... ' ....................... 37 Microchemistry: Availability and Needs ....................... 39 Emergency or Trauma Stations ................................. 46 Chemical Toxicology ............................................... 49 Interpretive ASPects ..... Analytical Aspects ........................................... Future Developments ................................. Hematology .............. ...... General ASpects ..... ............................ Instrumentation ........... Data Processing ....... .... Blood Bank ..................................................... 65 General Aspects ........................................... 65 Immunohematological Methods ............................... 65 Microbiology ........... . ....................................... 67 General Aspects .... ....................................... 67 Clinical Requirements and Feasibility ..................... 68 Serology .................................................. 69 Fetal Antigens and Cancer ................................. 69 Current Progress in Mechanization .............. .... ....... 7O Virology ........ ............. . .......... . ...... . ............... 75 General Aspects ... ..... . .................................. 75 Future Developments .... ..... . .................. .. ......... 75 Summary and Conclusions ........... . .. ...... . ... .. ...... FOREWORD The National Institute of General Medical Sciences is responsible for the Support of basic and applied research in the biomedical sciences and in a number of clinical disciplines, that are fundamental to the delivery of patient care. The Institute also administers a national program of research training through fellowships and research training grant programs to help meet our country's expanding manpower needs in the health field. Congress has strongly encouraged the National Institute of General Medical Sciences to support research and development in the field of clinical laboratory sciences. This decision was based on the realiza- tion that past and present systems of providing laboratory services to the public cannot be expected to cope with the tremendous increases in laboratory data and the precise determination of blood and body fluid alterations that are needed for the scientific diagnosis and treatment of the sick. The National Institute of General Medical Sciences established the Automation in the Medical Laboratory Sciences Review Committee as part of its concerted program of research and development designed to help further laboratory automation. The committee has the task of reviewing project applications and contract proposals for basic and applied research and developmental work designed to aid the automation of the clinical laboratory. The major areas of interest are clinical chemistry, hematology, microbiology, virology, and immunology. This report was prepared by the committee to aid the scientific community in identifying research and development projects that are needed in order to develop a completely automated clinical laboratory. It represents an assessment of the current status of the clinical laboratory sciences including the significance of new developments in the field and identified Specific needs and neglected areas in need of further development. I commend the committee for an excellent report. It will be valuable not only to the staff and programs of the National Institute of General Medical Sciences, but to many others who share their concern for delivery of accu- rate, rapid, and reliable laboratory results. DeWitt Stetten, Jr., M.D., Ph.D. Director, National Institute of General Medical Sciences CONTRIBUTORS Dr. Thomas D. Kinney (Chairman) Director of Medical Education Duke University School of Medicine Durham, North Carolina 27706 Dr. Elias Amador Associate Professor of Pathology Department of Pathology Case Western Reserve University Cleveland, Ohio 44106 Dr. Norman C. Anderson Director, Molecular Anatomy Program Oak Ridge National Laboratory Oak Ridge, Tennessee 37831 Dr. John L. Bethune Associate Professor of Biological Chemistry, Peter Bent Brigham Hospital Biophysics Research Laboratory Boston, Massachusetts 02115 Dr. James J. Conti Provost Polytechnic Institute of Brooklyn Brooklyn, New York 11201 Dr. Robert L. Dryer Professor Biochemistry Department of Biochemistry College of Medicine University of Iowa Iowa City, Iowa 52240 Dr. Merle A. Evenson Associate Professor Medicine and Director of Toxicology Laboratories University of Wisconsin Medical School Madison, Wisconsin 53706 Dr. Evan C. Horning Director, Institute for Lipid Research Baylor University Houston, Texas 77025 Dr. Robert S. Melville (Executive Secretary) Chief, Biochemical Sciences Section National Institute of General Medical Sciences, Bethesda, Maryland 20014 Dr. Morton F. Mason Professor, Forensic Medicine and Toxicology Southwestern Medical School University of Texas Dallas, Texas 75235 Dr. Morris Schaeffer Director, Bureau of Laboratories New York City Department of Health New York, New York 10016 Dr. Bert L. Vallee Professor of Biological Chemistry, Harvard Medical School and Director, Biophysics Research Laboratory Peter Bent Brigham HOSpital Boston, Massachusetts 02115 Dr. Arnold G. Ware Department of Clinical Chemistry Los Angeles County General Hospital Los Angeles, California 90033 Dr. George 2. Williams Director, George Zur Williams Research Institute of Laboratory Medicine Clay and Webster Streets San Francisco, California 94115 Dr. Theodore Williams Director of the Purdue Laboratory for Applied Industrial Control Purdue University Bloomington, Indiana 47401 Dr. Donald S. Young Chief, Clinical Chemistry Service Clinical Pathology Department Clinical Center National Institutes of Health Bethesda, Maryland 20014 THE CLINICAL LABORATORY The excellence of future medical practice will depend to a large degree on the development of new and improved laboratory measurements. The primary objective of the clinical laboratory is rapidly to provide accurate analytical data on samples obtained from human patients. By correlating these data with clinical information, physicians can make the best diagnosis or evaluate therapy for their patients with the least delay. The ideal situation for the physician is to have answers supplied as rapidly as he asks questions, as in history-taking or in measuring blood pressure. If decisions as to additional investigation or treatment of a patient are to depend on laboratory analyses, these results must be available to the physician in the shortest possible time so that he can promptly initiate a logical sequence of diagnostic or therapeutic measures. Laboratory data may be classified into two categories with respect to their significance in medical diagnosis: (1) determinative data that are essential to a definitive diagnosis, and (2) associative or near determinative data that are not unique for a specific disease but give confirmatory support to a clinical diagnosis. Determinative Data Data with determinative significance are usually applicable to disorders that are rare and impossible to diagnose Without laboratory confirmation. An example is chemical data in genetically based (missing enzyme) disorders. The existence of Refsum's Disease or Type II Familial Hypercholesterolemia, to name two examples, could not be confirmed without supportive chemical data. Analytical methods that produce Such determinative data are theoretically the most valuable, but they would not be generally available to practicing physicians because they are essentially research procedures, applicable only to rare disorders and hence economically impractical for use by hospital clinical chemistry laboratories. Most of the existing diagnostic analytical methods are usually as rare as the disease they are designed to identify, a situation that is not satisfactory for either clinical chemistry or for the practice of medicine. Associative and Near—Determinative Data Associative data are widely used in diagnosis of disorders that occur much more frequently than the missing enzyme metabolic disturbances mentioned above. A number of clinical entities are characterized by the loss of homeostatic control over the biochemical parameters that can be estimated with relative ease. Although the disorders themselves may be imperfectly understood, they have fairly recognizable clinical and physiological symptoms that can be confirmed by consideration of the appropriate biochemical data. For example, in uncontrolled diabetes the serum glucose values are high. If they are far higher than normal and if in addition the serum triglyceride values are much higher than normal, the problem of diagnosis is relatively easy. In patients with gout, the concentration of serum uric acid is generally higher than normal, and the diagnosis and treatment would probably not be undertaken without such chemical confirmation. Gout, however, can occur in patients whose serum uric acid concentration is only a little higher than the accepted normal values, and this illustrates the chief drawback of these data as a guide to medical action: the severity of the disorder in clinical terms may not be directly related to the degree of divergence from ”normal” for the biochemical parameters under study. At least three problems need to be explored: (l) The relations of biochemical parameters to disease states as shown by charting "high" and "low" distribution curves of single parameter data. (2) The associative value of different parameter measurements in Specific disease entities as shown by extensive computer-based analysis. (3) The possibility that frequent measurement of the same parameter on different occasions or many related parameters at the same time may be more valuable than the static measurement of a single parameter. A glucose tolerance is a primitive form of kinetic estimation method. Test procedures should be developed to challenge specific metabolic pathways. The first two problems have been studied by traditional methods for many years with but little effect on the clinical evaluation of such data. Instrument Requirements It is impossible to devise guidelines for an analytical machine that would be applicable to every situation in the clinical laboratory. Nonethe— less, it is possible to indicate certain essential characteristics of an analytical instrument for clinical laboratory use. The machine should be able to provide accurate and reproducible results, high volume capacity with a short delay time between sample entry and result, an accurate determination with the smallest possible sample, and a means of introducing emergency specimens into the system. Whether the analytical method is based on a chemical or a physical principle is not important so long as each step of a procedure can be identified sufficiently to allow simulation for programming and validation of purposes. Every step of a procedure must be monitored to detect any abnormal function. An alarm mechanism should be provided to alert operators to machine malfunctions, loss of specimens, and entering of erroneous results. Machines should function with a predetermined minimum of electronic noise and without drift. when a machine is controlled by means of feedback, the accumulated drift should not exceed a preset level. Above all, the machine should be reliable and require minimal maintenance. Wherever possible, modular constituents should be used so that the user can check and replace faulty parts with a minimum of ”down” time. One Of the mOSt pressing needs of the clinical laboratories_is a dependable current source of adequate and accurate information about laboratory instruments and equipment, including directions for their calibration. As a rule, personnel in the clinical laboratory have not been sufficiently critical consumers of commercially available scientific products. This has often resulted in continued use of substandard instruments that yield unreliable results. At the present time, many new instruments designed for the clinical laboratory have been inadequately field-tested before being placed on the market. As a consequence, there is often needless waste of money and time before such instruments can be put into operation, either because the instrument requires extensive adjustment or modification before it can be used or because it is incapable of performing the task for which it was purchased. With improved field evaluations, redesign and modifications of prototype laboratory instruments could be expected before commercial models are released to the general market. There is no doubt that more careful studies of prototypes eventually would provide more accurate laboratory determinations and decreased frustration for laboratory personnel. More attention should be given to safety precautions. There should be certification of safety specifications of all equipment, particularly in respect to grounding connections. At the present time, it is not uncommon for laboratory personnel to receive serious electrical shocks because of faulty equipment. To eliminate such hazards, new equipment must be validated for safety in terms of units as well as total packages. Because many hospital and laboratory accidents are the result of coupling instruments from different manufacturers without proper grounding procedures, certification of equipment should be required and should be awarded through a nonpartisan laboratory. Electric manufacturers currently maintain underwriter laboratories, but no such certification exists in the medical field. It would be extremely helpful if some method could be devised for a rapid feedback of information to both the manufacturer and the potential purchased. Possible components could include: (1) A registry of users, through which a prospective purchaser could contact the users for information. (2) An Evaluation Index which would use questionnaires would elicit information from users on the precision, down time, quality of service Supplied by manufacturer, etc. This information, in raw or in tabular form, would be available to interested persons. Manufacturers would be invited each year to update their replies to criticisms. (3) A series of ”users" conferences, at which laboratory scientists 9 who have used a specific instrument on a day—to-day basis meet with scientists, engineers, and manufacturers responsible for the development of the instrument to discuss its merits and shortcomings, exchange experiences in actual use, and offer suggestions for improvements in operation. Typical of the effectiveness of this approach were the successful conferences on the LING computer. Where there has been premature marketing of instruments with basically sound ideas requiring only further refinement, the users conference approach would serve to salvage the work and redirect its development along lines most useful in the clinical laboratory. (4) Comparison studies of different instruments purporting to do the same tasks by an unbiased organization that would make the results available to prospective purchasers, the objective being to stimulate the production and use of Successful instruments, and to encourage development of further systems. Sample Identification It is clear that the development of effective methods for positive identification of samples is a challenge of the highest priority. There is a great need for dependable automated systems that identify a specimen with a patient, the tests to be performed, and the results generated, and also facilitate data storage and retrieval. Such systems should have the following features: human and machine readability, accuracy, reliability, flexibility, low cost, convenience, and rapidity. In general, the clinical laboratory instrumentation currently in vogue and/or under development may be classified as using either continuous flow or discrete sample analyzers. While sample ”labeling,‘ in principle, 10 may be done either internally or externally, the latter method predominates. In external identification techniques, which utilize tags, cards, labels, charts, etc., to record all information relating to the Specimen, the "label” follows the sample by being directly attached to it or by tracking its progress with Subsequent matching. Postive and foolproof automated methods for controlling the flow of information associated with individual samples to replace error-prone manual and semi-automated techniques are required. To date, little effort has been devoted to the satisfactory development of integrated instrument Systems that include automatic sample identification as a component of the overall design. It is likely that progress for the short range will be exhibited in the mechanization of the information processing asPects of existent and emerging instrument systems. A universal automated sample identification scheme is not available because of the broad spectrum of tests and the widely varying sample volumes involved. Never— theless, there are some applications in which progress may be stimulated by a re—evaluation of present—day sampling techniques and specimen containers. An overall goal is to handle the specimens in such a manner that there is no possible way they can be mislabeled. This means that provision must be made for specimen identification at the bedside by some method that can be used in the laboratory to insert the results in the patient's record. To obtain maximum reliability, information would include the identification number and, ideally, the patient's social security number and name. Consideration must also be given to using a technique of identification that will fit into the hospital's system of bookkeeping and patient charges. Input devices for sample identification data and for reSults of 11 manually performed tests at present include typewriters, mark-sense and punched card readers, and Specially designed keyboard consoles for specific purposes. Special keyboards coded for cell types, one used in differential cell identification (microscopic) and enumeration for peripheral blood and another for bone marrow examinations are in daily use and are available commercially. Data and information output are accomplished by a variety of devices including typewriters and high-speed printers. Typewriters are slow and quickly produce a bottleneck in even the smaller laboratories. High-speed printers are expensive but efficient, and permit more rapid production of reports, work lists, and quality control data. Cathode ray tubes are used eXperimentally in a few developmental systems, but require sophisticated and memory-core-conSuming computer programs. These are not yet of proven value, but offer promise as quality control and error feedback devices for the laboratory. One or two systems make use of multiple typewriters, one in each of several hospital wards for patient-test report transmi351on. These are noisy and slow. They do not permit review and evaluation by the laboratory professional staff and they must be operated by ward clerks and nurses. The reports produced are usually only partial, and they are difficult to organize and read. Printed reports have the advantage of high-speed production, easy review by laboratory physicians of any abnormal results obtained, and an easily readable format. Processing programs vary from elementary sorting and listing of final results as transmitted to the computer by input devices (cards, typewriters) to more sophisticated programs that receive and collate 12 patient, sample, and identification data as well as raw test reSults. These programs provide for appropriate computations on the raw data and for comparison with similar data previously obtained on the same patient. They also permit collation and print out all test results of a given patient for a designated period. More complete programs can also correct raw data for assay errors owing to instrumental drift and the like. Most of the above programs are specific for the laboratory situation in which they were developed, but a few are commercially available as part of a system. Program software is written for a specific configuration of computer and related peripheral hardware, laboratory instruments, assay methods, and reporting needs. Such an approach is difficult to adapt to other systems. The commercially available programs operate with their related hardware as a system, and require adaptation of the laboratory operation to the constraints of the selected system. Yet to be developed are modular subroutine program packages for use with hardware modules with options for type and size of operation involved. The solution of this problem would seem to hinge on the application of systems engineering techniques. It will require the involvement of clinical chemistry research laboratories, instrument manufacturers, and manufacturers of data-processing systems. Data Reduction A large and promising area needing develOpment is how to use laboratory data most effectively. General and widely useful computer programs are needed; for biological or physiological validation of specific laboratory data of patients, for correlating groups of related data (including clinical 13 information), and for the generation of rational physiological concepts and management pointers oriented to the specific patient. A sharp distinction should be made between data manipulation and data reduction to information. The terms are self-explanatory and one should realize that most efforts to date have been in the area of da£§_manipulation. Data available from modern comprehensive laboratory analyses are so complex as to bewilder even the most knowledgeable physician. This leads quickly to the habit of limiting laboratory examinations to those items that can be quickly understood and used. Physicians are expected to remember ”normal” ranges for hundreds of tests. Frequently, with changes in methodology, ”normal" ranges are modified. Some enzyme measurements have changed units as many as three times in the last five years. Most of the effort of the last five years has been spent on the development of data manipulation systems. Many workers have concentrated on a small area of laboratory reSponsibilities such as hematology and clinical chemistry. A few have worked on overall laboratory systems, and some have attempted to develop complete hospital information systems, but a complete, well-developed system for all areas of the laboratory is still not available. As more and more human ills come to be understood, at least in terms of their biochemistry, a large battery of sophisticated tests will be required. Recognition of many minor genetic diseases or predispositions will give rise to genetic typing of great complexity. As a consequence, even a Specialized physician will not know all the normal values for all tests or the implications of all deviations, especially of clusters of 14 inter—related factors. Experience has shown that no one individual will be able to draw meaningful conclusions from the mass of data, to express it in terms of probabilities, and to indicate what additional tests are needed. Thus, the potential value of readily obtainable data cannot be realized with the data-handling methods now being used. Complete mastery of very complex computer-analytical instrument systems will be required if the following questions are to be answered: (1) How can large numbers of new tests be evaluated statistically on large numbers of patients? (2) How should they be selectively used? (3) How can one meaSure the predictive value of tests? (4) How can we discover if the results are properly interpreted and used? Broader questions that must be considered are: How can data, on humans largely from medical records, be processed in such a manner as to allow conclusions to be reached? How can we identify and document new problems? How can we objectively assess the benefits of new developments? If a new environmental insult produces measurable change in biochemical parameters in a large segment of the population, will this be noted? If a new therapy produces one or more unexpected small alterations in blood constituents, will this trend be winnowed out from the mass of data? How will the advantages of more rapid or more precise analyses be demonstrated? Medical records provide a wealth of data, both on the course of individual cases, and on population trends, which are not always available for computer processing. This complex problem has both technical and ethical difficulties, but, to assess both individual and population health trends the problem must be solved. 15 Computer Emulation It is often desirable to convert programs written for one manufacturer's computer to a form that can be used for a different make or to produce programs on one computer for Specific use on another. For example: (1) Special programs have been written for the IBM System 360 series computers that allow them to run programs previously written for the IBM 1400 series machines. Only by the use of these special programs was IBM able to promote the rapid transition of business organizations to the 360 from previous models. In this case, the effort was carried out completely by hardware or directly machine wired programs. Such a program is called an "emulator.” (2) A large computer can be used with a major set of programming aids to prepare programs for a small computer that has only very limited memory and without these programming aids in its own repertorie. In this way it is possible to program small computers with only minimal memories by use of relatively easy techniques, such as FORTRAN or BASIC compilers, that would require a much larger system if implemented on the small computer alone. An example of this is the use of the General Electric Company's remote time-sharing system computer as the medium for programming the Digital Equipment Corporation‘s PDP-8 series machines, a combined operation that was carried out by the Electro Scientific Industries Company of Portland, Oregon, for their line of computer controlled electronic test instruments. Business system emulator programs Such as those discussed in Item (1) have been and are being extensively used; it is estimated that fully 16 half of the use of the System 360's now installed is mainly in the emulation mode. A large proportion of these machines will probably never have their work converted to the more efficient programs specifically written for the new machines because of a lack of the manpower to carry out the necessary programming work. This manpower requirement exists in the small computer application field to an even greater extent, because these computers are being extended into areas in which they have not been used before and in which a residual of application and programming skills does not already exist. In addition, there is the great economic incentive to keep the machines used for these tasks as small and inexpensive as possible. This will again make their application easier in the commercial rather than the engineering sense. Programming in a compiler level language is much easier both to learn and to perform, because the computer itself performs much of the clerical-type work involved in the programs. At the same time, it handles most of the involved tasks, such as memory assignments and program linkages. Unfortunately, the programs or sets of machine instructions that make possible easier writing of further applications programs are themselves very lengthy. They will thus require major amounts of memory and of computing capability on the part of the same computer if it is to compile its own applications programs. This would, of course, negate our desire for combining programmingeease with small, inexpensive machines. The emulator principle, however, may provide a solution to this problem. Large scientific and business computer systems are normally l7 equipped with memory capacities that are more than adequate for handling desired compiler systems. We must then produce an additional program, the emulator program, written for the large computer, which will allow it to modify the operation of its own computer system in order to produce an object code or final program for the desired application which is recognizable and usable by the small computer to be used on thisexperiment. The scientist can then do all his applications program preparation on the large machine with the assurance that the small machine can use the resulting work. As mentioned in Item (2), this type of emulator already exists to permit the General Electric GE 235 computer to produce programs for the PDP—8, 8% and 81 computers. They also exist for several other combinations of large and small computers or can be readily written when the need arises. Miniaturization Technological advances in computers, sensors, and in electrical components such as motors, switches, controls, data displays, and in materials and design make possible the construction of instruments that are smaller than those presently in use. A number of the advantages that could result from miniaturization are: (1) Space saving. A large fraction of both bench and floor space in a modern research or clinical laboratory is taken up with equipment. This is especially true in a research laboratory concerned with the development of a new mechanized or automated system for clinical research. In many instances, more than 50 percent of bench space is occupied with spectrophotometers, pH meters, continuous-flow analyzers, cell counters, and similar devices. If the object of constructing laboratories or 18 laboratory buildings is to obtain working (or bench) space, then the total cost of providing one linear foot of uncluttered usable bench-top is very high indeed . (2) Ease of standardization. Miniaturization has always resulted in an increase in the number of components integrated into one replaceable part. Operational amplifiers, integrated circuits, and printed circuit boards are cases in point. As larger circuits and systems are miniaturized, an instrument finally becomes a manageable collection of functional modules which, if properly designed, can be checked by a relatively inexperienced operator and replaced by him if found defective. The best example of this trend is found in Nuclear Instrumentation Modular Systems (NIMS) developed jointly by the AEC and the National Bureau of Standards. A complete counter can be assembled from preamplifiers, amplifiers, scalers, etc., frOm different manufacturers. Although new developments in electronics will continue, reliability of components is such that laboratory instruments should last a long time. By insisting on standardization of components, any possible tendency of manufacturers to build in obsolescence is countered because reliability and performance are readily intercompared by the user. For the clinical laboratory, pH meter and spectrophotometer circuits could be standardized easily, as could the amplifiers associated with (e.g.) recording potentiometers. (3) Maintenance. Repairs can be minimized and simplified for the user by combining miniaturization with a decrease in the number of replaceable parts. Amino acid analyzers, for example, are presently repaired in the field, a costly and time consuming process. If the space occupied by Such a machine can be reduced to one or two cubic feet, it will 19 be more efficient to move the instrument to the repair specialists rather than vice versa. When problems arise, a machine in operating condition is dispatched at once to the user, who returns the defective one in the same container. (4) Reagent requirements. If labor costs are excluded, reagent costs in the clinical laboratory constitute a large fraction of the assay costs. It is estimated that for each dollar spent on instruments, one dollar is spent each succeeding year for reagents. In some instances, instrumentation has been supplied free providing the user agrees to buy all the reagents for it from the manufacturer. By reducing sample and reagent volumes by a factor of 10, it is evident that tens of millions of dollars could be saved. (5) Egg}, Standardized miniaturized systems should ultimately be less expensive than present large systems because of ease and low cost of maintenance. Emphasis will increasingly be on the measurement and transfer of small fluid volumes, and on mixing and making measurements on small volumes (l-20}ul range). The problems have to do with forces used to move such volumes; with the nature of the forces that ensure or prevent the wetting of Surfaces; with mixing, and with measurement. This is in some reSpects a new field, and the importance of it should be more widely understood. Absorbance measurements with such small samples require special optical systems, methods for eliminating bubbles and stray particles, and ways of cleaning small cuvettes if they are to be reused. If electrodes are to be used for measurement, drift, calibration,and specificity are problems. Where enzymes are used as reagents or as components of 20 electrodes, problems of enzyme stability or of enzyme poisoning require careful consideration. Quite obviously, much additional work needs to be done on the chemistry of many analytical reactions. Emphasis should be on performing analyses kinetically. This eliminates many problems associated with blanks and with some interfering substances. Cost Effectiveness Documentation of well-defined cost benefits is scant and applies only to partial systems or to individually automated tests. Available information, however, indicates that in large laboratory operations, automation and data processing are economically effective in decreasing operations costsl. Several careful, detailed studies of the cost benefits and service effectiveness of automation in small- and medium-sized laboratories are needed as guidelines for future development. It must be recognized more- over that effectiveness in service is of equal or greater importance than cost effectiveness. For instance, if automation and data processing were to improve the accuracy and reproducibility of 25 percent of all tests performed and if errors in results are decreased from eight to ten percent to one or two percent, the improvement in quality of service and benefit to patients would be great but could not be calculated in terms of actual dollars. Although specific documentation is not yet available, the experience of knowledgeable laboratory directors in handling increasing workloads with reduced numbers of technical personnel indicates that in larger lG. Brecher and H. F. Loken, ”The Laboratory Computer--Is It Worth the Price?" Am. J. Clin. Path. 553527, 1971. 21 institutions with annual laboratory budgets approaching one million dollars or more, automation is effective and can be afforded at costs of $150,000 to $200,000 per year. In smaller laboratories,those operating on a yearly budget of less than $500,000 per year cost effectiveness of computer automation at current prices is questionable. Useful Systems costing $20,000 to $75,000 per year in leased equipment and computer time, however, would be tolerable financially and could increase production and efficiency. At this time, most of the mechanized laboratory equipment has been designed to produce large volumes of determinations in a short period of time and is not practical for use in the smaller laboratories. Nevertheless, there is a need for mechanized devices that will provide” laboratory data accurately and quickly for these small laboratories and not require the depth of knowledge needed to run the more complicated large equipment. CLINICAL CHEMISTRY Methodology The goals for the automation of clinical chemistry should be to extend its usefulness, for example: maintaining a high level of accuracy and precision; providing multiple analyses on small samples (lO/ul or less) of blood; delivering rapid performance and immediate transmittal of results to the physician, and assuring maximum reliability 24 hours every day at an acceptable cost. New and improved procedures must be developed to exploit the latest advances in modern scientific and engineering technology. A starting point may be found in studies made in research laboratories, or in clinical observations which lead to a new clinical laboratory procedure. The ”reduction to practice”'Gnethods suitable for mechanizing or automating large numbers of samples) stage is usually reached when there is agreement between laboratory scientists, who decide that a given procedure is satisfactory in a scientific sense, and c1inicians,who regard the resulting data as useful in the diagnosis and treatment of disease. There are functional relationships between the development of new methods, their application in the clinical laboratory, automation and bench evaluation. (Figure 1) Development of New Procedures / Development \ .____________1 \ Evaluation I " Automation l \ Application / The establishment of a new clinical chemical procedure which involves the least time between inception of ideas and ”reduction to practicell must be preceded by a testing period during which new areas of automation are evaluated as to usefulness. When there is agreement that a given item of information has value in clinical work and when it can be obtained by one or more methods, an effort is usually made to provide the information more economically and efficiently. At the present time, Such effort usually provides "automated" procedures that require very little human action or supervision except for setting up and maintaining the system and for arranging for sample introduction. The importance of studying and developing procedures for sample or reagent transfer is obvious; the methods may be modified almost indefinitely to meet the requirements of different procedures. Any advance in automation methods carries with it potential usefulness for many applications. This potential usefulness, however, depends also upon the state of competitive methods based upon different kinds of procedures. For example, at the present time object transfer within a system is the oldest of the methods for ”assembly line" analytical or other procedures. Object transfer systems are relatively easy to devise, but they are also for a variety of reasons the least useful of all present day analytical systems. Advances in methods which involve either liquid transfer or gas transfer within a system are more likely to affect the future of clinical chemistry. So—called new methods of analysis uSually evolve from experimental testing of theoretical methods or of existing methodology of unrecognized significance. For example, chromatographic absorption and partition 24 separation methods were described many years before their general usefulness was recognized; the establishment of modern chromatographic theory and practice has been a relatively recent but highly important advance in chemical work. The stimuli for new methods usually have come not from clinical chemistry but from research laboratories. Studies of special applications (for example, methods for use in emergency stations or for newborn infants) have generally received very little consideration except in conventional terms. Review of currently available methods and desirable new developments reveals the following categories of functional or operational features common to each group: A. Methods for "automating” established analytical procedures 1) involving liquid transfer within a system 2) involving gas transfer within a system 3) involving solid or object transfer within a system B. Development of improved separation or discrimination techniques and new identification or estimation techniques 1) chromatographic separation methods 2) electrophoretic separation methods 3) molecular size separation methods 4) other separation methods 5) discrimination methods: enzymic activity 6) radio-immune assay 7) other techniques A. Methods for "automating" established analytical procedures 1) Methods involving transfer of liquids within a system Several methods of liquid transfer have the characteristics needed for automating analytical procedures depending on chemical reactions occurring in the liquid phase. These include the segmented flow procedure used in Technicon systems and the newer GeMBAEC analytical system. The GeMSAEC system offers the possibility of making one or more measurements at any time during the course of a reaction. Other methods of liquid transfer within a system could be developed. The requirements for movement of a liquid without mixing (until mixing is desired) can be met in other ways. The Dupont analytical instrument, for example, is designed to transfer liquids through successive compartments in a single package. The chief limitations for all liquid phase analytical methods are uSually those imposed by the nature of the detection systems which are available. A less serious limitation is that multicomponent analyses are usually slow because of the inherent slowness of liquid phase separation (chromatographic) systems unless run under high pressure; this will be discussed later. Liquid analytical systems (as distinguished from gas-liquid systems) for use in the clinical laboratory may be classified as: Class A: Analytical systems in which a multiplicity of samples is analyzed for a single substance using one specific reaction. Included are: a) Discrete analytical systems in which samples and reagents are diSpensed volumetrically into a series of separate containers or vessels. 26 b) Serial discrete systems in which one sample after another passes a series of process stations and where the final measurements are made serially. c) Parallel discrete systems in which a group of samples are all subjected to the same procedure at the same or almost the same time. d) Continuous-flow analytical systems. This group includes the Autoanalyzer and a few other systems, some of which are still in early development. Class B: Analytical systems in which analysis depends on a separations method with detection of the separated molecules being done by a relatively nonspecific method (refractive index, ultraviolet absorption, conductivity, etc.) or a group specific method (ninhydrin reaction, anthrone reaction for sugars, etc.). Included are the following methods: 1. Chromatography 2. Gel Filtration 3. Electrophoresis 4. Sedimentation analysis The Class B systems, with the exception of cellulose acetate electro- phoresis and a few chromatographic gel filtration columns, are not routinely used. The chromatographic systems which would be of general interest (Such as the amino acid analyzer, the nucleotide or UV analyzer, and the sugar analyzerx at the present timq are too expensive and are too slow for routine use. Analytical ultracentrifugation is likewise rarely available on a routine basis in the clinical laboratory for the same reasons. 27 In Class A l, a large number of new systemsare available, but little,comparative,experimental data are available for most of them. It is, therefore, not possible to state how many of these devices will prove to be successful in actual practice. The parallel array of systems using GeMSAEC principles are also too new to be properly evaluated at this time. The success of the Technicon Autoanalyzer (Class A) is well known,and for the present most new systems will be compared with it. 2) Methods involving transfer of gases within a system The ordinary gas-liquid chromatographic system is a gas-tight assembly which operates under positive preSSure at the front, or sample-introduction end, and at atmospheric pressure at the discharge (after detection) end. The sample, which is usually a solution containing solvents, reagents and unwanted "impurities" in addition to the substances under study, is introduced and vaporized in an ”injector" zone. The compounds under study traverse the chromatographic system and sequentially enter a detector device. The solvents, reagents and all volatile constituents of the sample follow the same route. A mechanized or automated system, to be fully effective, should have provision for venting unwanted solvents and reagents in Such a way that these materials donottravel through the same analytical column as the sample. Provision should also be made for transfer of part or all of a sample from one type of column to another. Any operation in a system which involves sample transfer requires the use of low-volume valves. Until recently, these were not fabricated from materials that have a 28 long life at elevated temperature. The sample which is introduced into the chromatographic system may be in solid, liquid, or gaseous form. The analysis of gaseous samples is usually employed for special problems, such as process control sampling of gas streams, breath analysis, and head Space analysis. Solid and liquid samples are usually introduced in solvent or reagent mixtures. The usual sample size is l-lOIJl, and introduction into the system is by syringe. In order to permit automatic or mechanized sample addition, devices have been developed in England and the United States for inserting metalgauaze supports or Teflon tubes into a GLC column. when metal Supports are used, the samples are introduced in a solution 'that is evaporated to yield a coating of the sample on the metal gauze. This method of sample addition is not widely used, chiefly because of problems of stability of derivatives under these conditions. Recent developments include an automatic syringe injection device, and an automatic sampler that employs aluminum or gold capsules that are pierced after being positioned in the inlet system. If the instrument can be constructed that will permit the automatic transfer of gas samples from one column to another, the next step would be to develop a fully automatic sample addition system. Since the phase transition is usually best accomplished inside the system, this should take the form of a device for injecting liquid samples. In order to avoid opening the system for the removal of supports or containers, the sample should be injected into a vaporizing zone. GLC systems are now available in a form which includes automatic cycling. The temperature of the chromatographic oven can be lowered 29 to accept a sample for a temperature programmed separation; after the separation, the cycle can be repeated. This equipment requires very little modification for automated operation with transfer of gas samples. 3) Methods involving solid or object transfer within a system are ”assembly line" methods. They are of historical interest but necessary in some analytical work. For example, in nuclear activation analysis, objects must be transferred from one station to another by automatic methods. B. Development of improved separation or discrimination techniques and new identification or estimation techniques l) Chromatographic separations methods are generally classified according to the separation process: liquid—liquid partition chromatography, gas-liquid partition chromatography, liquid-solid absorption chromatography, and mixed systems which involve both partition and absorption. In this discussion, ion exchange methods are included as well as the closely related counter-current distribution methods. It is obvious that improvements in separation procedures will provide options that may prove to be useful in devising clinical chemistry pro- cedures. Research on development, however, will not be of immediate or near-term value unless the methods can be quickly moved to automation. (Figure 1) For this reason, developmental work shOuld be evaluated both as to this possibility, and as to the type of information that may be generated. In general, the time involved in moving from this area to "reduction to practice” in clinical chemistry will probably be several years. 30 Two types of systems should be considered. These are based on (1) nature of the phases, and(2) effectiveness of the separation. Current practice for column separations is to use long, narrow columns with flow-through detectors, with devices for sample admission and for recording of responses. The trend in liquid phase systems is toward longer columns (resembline gas phase columns) and toward higher preSSures of operation. Gas-liquid chromatographic systems are still commonly operated at lower pressure (to about 50 psi) using packed or capillary columns. Many different detection devices can be used in GLC work; comparatively few are available for liquid phase systems. Paper and thin-layer systems are useful in many laboratory applications, but they are not well suited for ”automated systems." Ion exchange chromatography may be regarded for present purposes as a form of liquid phase chromatography. Paper, thin-layer and many liquid phase column methods are low resolution procedures. Moderate resolution, however, can be achieved by liquid-liquid, liquid—solid, and gas-liquid column systems. High resolution is at present achieved only in GLC systems, although good separations may be achieved with low resolution methods by appropriate choice of conditions. For example, TLC systems have low resolution (band spreading);but with the proper liquid phases, good separations are usually possible. The number of compounds that can be separated at the same time in a low resolution system is usually low. Only high resolution systems will separate many different compounds in a single run. There are several problems that should be investigated:(l) the 31 design of GLC systems which may be ”automated;”(2) the design of high- pressure liquid-liquid systems which may be "automated" (the major problem here is the lack of variety of available detection systems); and(3) an investigation of ”dense gas” chromatography. Application is typical of work that should be supported in the developmental area. The theoretical aspects are not as well-defined as in other forms of chromatography,but the preliminary experiments have been promising and the limitations at present are largely those of instrument design. 2) Electrophoretic separation methods are useful in several applications in clinical chemistry. The techniques are well established, and the chief problem has been that of developing reliable instruments requiring a minimum of hand work. 3) Molecular size separation methods such as gel permeation techniques are not widely used in clinical chemistry at the present time. Further development may be necessary before these methods will be useful. 4) A number of other separation methods are theoretically possible, but have not been evaluated or are not available in the form of experimental systems. For example, the ”Plasma Chromatograph” resembles a mass Spectrometer in that charged ions are separated in a gas phase, but the ions are generated by ion-molecule reactions and the separation chamber is at atmospheric pressure. Gas centrifugation can be used to separate neutral molecules of different mass, but the principle has not been evaluated for biochemical separations. 5) Discrimination methodg such as those based on enzymatic 32 reactions possess potentially high specificity. Biological samples obtained from hospital patients, however, may contain drugs or metabolities which may adversely affect the reactions without being apparent. Most current procedures involve optical measurements for determining concentrations and are based on well established methods. Other detection systems based on fluorescence, electro- chemical methods, and isotopic labeling have potential usefulness although these procedures are not much used in the clinical laboratory at this time. The development of techniques for binding enzymes to resins and other media facilitate the analyses of specific substrates, e.g., glucose and urea. Further application of these techniques will allow simple and accurate measurements of additional compounds,such as lactate and pyruvate. It is probable that enzymes from microbiological sources can be obtained,but there are as yet no simple specific procedures for this purpose. The measurement of enzyme activities as indices of tissue damage has had considerable application in the clinical laboratory although many measurements are often made without regard to optimum concentrations of substrate, cofactor, and secondary or coupled enzymes. In many instances, the pH, temperature, and buffer conditions are not optimal. Much existing equipment necessitates the use of a single end-point determination, making it impossible to use kinetic measurements. With kinetic measurements,a shorter reaction time may be used and fewer problems from side reactions occur. Human enzyme reference materials should be developed for 33 verification of enzymatic procedures that are now dependent on enzymes of animal origin. Many enzymes measured in the clinical laboratory lack adequate tissue specificity; a determined effort should be made to search for organ-Specific enzymes, or isoenzymes, in order to achieve greater certainty in diagnosis. 6) One of the most promising new methods is based on the use of stereospecific absorbance to detect and meaSure trace quantities of hormones and a variety of antigens. In radio—immunological techniques, antibodies or similar specific molecules are used to bind antigens labelled with a radioisotope. When such complexes are mixed with solutions containing very small quantities of the same antigen or antibody in unlabelled form, an equilibration takes place and a portion of the labelled antigen or antibody is released into solution where it can be separated and counted. Drugs and hormones are highly reactive substances which are bound on specific receptor sites in cells. These receptors may be isolated and used in an analogous manner. The method has wide applicability but requires much additional research and development,which should be enCOuraged. 7) Several other physical-chemical measurements are applicable to clinical chemistry. Infrared spectrophotometry has been used for kidney and gall stone analysis and for preliminary drug identification of the purified sample. Its usefulness in drug analysis has been limited and this technique is now largely replaced by gas-liquid chromatography. The mass spectrometer has great potential usefulness in the clinical chemistry laboratory. The interfacing of the mass spectrometer to gas-liquid chromatography has been expensive but scientifically 34 successful. Current work is directed toward developing "analytical systems" which combine a gas chromatograph, a mass spectrometer, and a computer. Atomic absorption Spectrophotometry has become the analytical method of choice for calcium, magnesium, copper, and zinc in the clinical laboratory. High sensitivity for a large number of metals in combination with minimal chemical and optical interference make its application to clinical chemistry useful despite present limitations in precision. The use of emission Spectroscopy for trace metal analysis in the clinical laboratory has been limited due to the extensive sample preparation necessary and the lack of quantitative, rapid,and inexpensive readout of answers. Emission spectroscopy has adequate sensitivity for a wide variety of metals, but to date operational difficulties have restricted its use. Circular dichroism and optical rotatory dispersion are two measurements employing circular polarized light to study molecular symmetry changes of biologically significant molecules. Protein and enzyme structures, binding and interactions have been studied most frequently by these instrumental approaches. Neutron activation analysis has found only limited use in the clinical chemistry laboratory. To date, chemical separations of highly radioactive samples have been necessary before gamma ray Spectrometry and counting could be performed. In the field of electroanalytical chemistry, several techniques have been used in the clinical laboratory. The potentiometric measurements using ion selective electrodes have made significant progress recently. Measurements of the dissolved gases, oxygen and carbon dioxide, are commonly performed electrochemically. X—ray diffraction, X-ray fluorescence,and X-ray spectrometry have found limited use in clinical laboratory analyses in the areas of concretion and analysis and trace metal determinations. Progress in techniques and expanding use of nuclear magnetic resonance and electron spin resonance methods can be expected in clinical chemistry measurements. C. Special Problems 1) Detection of Metabolic Disorders At present, two approaches are available for the detection of metabolic disorders: @) Specific identification of a chemical constituent of a body fluid for a single disorder, e.g., phenylketonuria, and;(b) the nonspecific detection of many disorders. Two methods have been widely used:(l) incubation of patient sample with specific bacteria whose growth is dependent on the presence of particular amino acids, e.g., as used in the Guthrie test, and(2) Specific chemical determinations, e.g., of phenylalanine by fluorometry. Mechanized equipment is commercially available for analysis of phenylalanine in blood by both these means. The generalized approach to the detection of metabolic errors involves chromatography of blood or urine. Chromatography on paper or other thin—media makes it possible to study many samples simultaneously. Column chromatography (ion-exchange or gas-liquid) enables single samples to be studied in greater depth, but this technique as presently used is slower than paper or thin-layer chromatography. These latter techniques are suitable for detecting disorders of amino acid or carbohydrate 36 fi—_fi metabolism but are less suitable for investigation of other forms of disordered metabolism, e.g., of lipids or purines. Many devices have been manufactured to permit quantitative addition of samples to papers or thin-layers, and densitometers are now available for direct quantification (though this may not be as accurate as desirable) on the medium. No equipment is available commercially at this time for mechanizing chromatography on thin-media. Apparatus devised by Bush2, however, does have the potential for performing all the operations necessary for mechanizing paper chromatography. Tocci3 has also disclosed preliminary information on an apparatus which may partially mechanize thin-layer chromatography. 2) MeaSurement of Therapeutic Drugs in Physiological Fluids There is likely to be an increasing need for measurement of serum concentrations of drugs and their metabolites in order to determine correct dosage schedules. Few drug concentrations are now determined because of technical difficulties; yet optimum dosage for each patient should be assessed from the concentration of the active principle of the drug in his blood. 3) Tissue Analyses There are a number of situations in which tissue analyses are being used at present. Certain metabolic disorders may be confirmed from the 2 I. E. Bush, ”Automation of the Analysis of Urinary Steroids Using Quantitative Paper Chromatography and a Small Laboratory Digital Computer." Clin. Chem. 14:491-512, 1968. P. M. Tocci, ”A New Screening Method for Inborn Errors of Metabolism.‘ Clin. Chem. 15:795, 1969. l recognition of enzyme deficiencies in leukocytes or fibroblasts. The measurement of vitamins in blood cells gives an index of tissue saturation that is more satisfactory than their measurement in serum or urine. Tissue samples obtained by biopsy may be used for determina- tion of specific enzyme activities. Most hospital laboratories are not prepared to do this, but such procedures should be anticipated as potentially routine for large hospital centers. Lack of Suitable instrumentation delays their establishment as routine procedures. D. Microchemistry: Availability and Needs In most general hospitals, fewer chemical tests are performed on the blood of an infant than on the blood of an adult with the same disease. Assuming that only those tests which are actually required for good care of the adult patient are being performed, the child by inference must be receiving less than optimum care. There are several reasons. The collection of blood from infants is more difficult than from adults. It is often difficult to obtain a sample which is suitable for analysis, and analytical procedures on the small volumes obtainable from infants require greater technical skill than those for the larger samples obtainable from adults. Most general laboratories are oriented to the analysis of large numbers of samples in one batch by mechanized equipment for use with adult volumes of venous blood and are not able to process efficiently the infrequent sample from the pediatric ward. Manual microchemical analyses require a much greater investment of personnel time than the corresponding tests performed mechanically; consequently,they cost more. For this discussion, the maximum desirable sample volume for one 38 test on an infant's blood is regarded as 25}ul though 1 to leul has been advocated by many specialized pediatric laboratories. This volume would permit multiple tests on capillary blood collections. The tests performed on infants are the same as those for adults, with the addition of special procedures for inherited and congenital disorders of metaboliSm. The factors essential for good quality results from the laboratory are similar for both adults and children. Initially, a blood sample must be obtained which is both adequate in volume and quality. Venipunctures are impractical in young infants, so capillary blood obtained by heel- prick or other means must be used. For chemical analysis of any significance, reproducible samples must be obtained, uncontaminated by lymph, tiSSue fluid, and debris. Too much variability now exists between replicate capillary blood samples obtained by different physicians and technologists. There is a need for simple equipment to obtain samples which, if used in a prescribed manner, would eliminate the variability due to differences in collection techniques. The Small volumes of sample, which are obtained and collected into small containers,present Special problems of maintaining patient identity during processing because it is impossible to attach a label or write directly on the container. Even a simple oral verification of the patient's identity as the sample is drawn is difficult. Since most hospitals aSSure the identity of patients, pediatric in particular, by a physically attached identification tag, mechanical transfer of this information to the container is possible. Such identification could also be read by analytical instruments in the laboratory. The requirements of specificity, accuracy, precision, and sensitivity 39 of methods are, in general, greater for analyses on children's blood than for that of adults. For example, because of the different dietary composition of Sugars and the less well developed intestinal enzyme systems, procedures for blood sugar determinations must measure true glucose only, with ancillary methods available for the determination of other sugars if necessary. The increasing use of drugs, structurally similar to analytes determined in the clinical laboratory, also imposes extra requirements of specificity. These problems have not been recognized in the past for tests on blood either from adults or from infants. Inaccuracies in pipetted volume of samples are probably the largest source of errors in analyses on small volumes. The precision of most mechanized pipettes is poorest when the sample volume is small. Pipetting by hand is also more prone to error for small volumes. There is a much greater need to ensure quantitative delivery of small samples with non-wetting surfaces and absolute cleanliness of equipment. In most but not all cases, it is possible to scale down the analytical procedures for adult-sized samples; but for small samples precision of all ”hardware” involved,such as colorimeters or spectrophotometerg may have to be improved. The mixing of small volumes of solutions require special devices. Due to limitations of current techniques and instrumentation, requirements for sensitivity of analyses that would permit a reduction in sample volume have not yet been explored. If existing variables in quality of sample and analytical techniques were overcome, increased precision would enable trends in concentration of blood constituents to be determined earlier, and hence permit rapid institution of corrective therapy. 40 Requirements for instrumental techniques in the conventional chemical analysis of urine constituents in infants are not as great as for blood constituents. The sample volume is largen and the equipment used for analysis of chemical constituents in serum or urine from adults is applicable. Devices are currently available for the quantitative collection of urine from infants, but it is impractical to relate quantities of constituents excreted to a fixed collection interval. To some extent, this reduces the requirements for precise analyses of urinary constituents. There is a need for apparatus to permit semi- quantitative screening for abnormal constituents in urine and other fluids by simplified low-cost procedures. The available volume of amniotic fluid is usually sufficent to permit tests to be performed with established routine procedures. The difficulty of collection of sweat, saliva, and other secretions necessitates the use of procedures similar to those used for pediatric blood analyses. 1. Chemical Methods Natelson4 has reviewed the principles basic to microanalysis. In Summary, these requirements are: 0) conformity with existing procedures with scaling down of all volumes used in the regular procedure with the same sequence of addition of reagents;(2) use of procedures with minimum number of transfer operations, e.g., pipetting and filtration;(3) use of procedures for which high purity standards are available; and @0 use of procedures which develop a stable color of high absorptivity and adhere S. Natelson, Microtechniques of Clinical Chemistry, (2nd Ed.). Thomas, Inc., Springfield, 111., 1961. 41 to Beer's Law. The first requirement should now be modified to permit the use of superior procedures,which were first introduced for analysis at the lO/ul level, e.g., the Beckman ERA-2001 glucose analyzer employing a polarographic oxygen sensor to measure the rate of utilization of oxygen from the action of glucose oxidase on glucose. By adhering to these principles, it is possible to achieve a level of reliability of chemical analyses on a few pediatric samples at a time comparable to that achieved when larger volumes are used, even when the latter are mechanized. Precision deteriorates, however, when large numbers of analyses have to be performed manually. Urea nitrogen and glucose analyses may be made colorimetrically on 1 or ZJul serum using specific enzymatic techniques. The volume could be further reduced if accurate pipettes were available to handle smaller volumes or special purpose spectrophotometers were available with minimal dead space in the tubing and Superior washout characteristics of cuvettes to minimize contamination of one sample by another. The use of fluorometry would permit a further reduction of sample volume if required and is certainly applicable for other procedures, e.g., bilirubin, for which the sample volume now required in most pediatric laboratories is greater than 25/ul. Theoretically, both spectrophotometric or fluorometric equipment currently available could be used for existing procedures if the equipment and procedures were modified to handle small volumes. Flame photometric analyses, both flame emission and atomic absorption, can be made on sample volumes less than 25)Jl because of the high dilutions used. Instrumentation is available to make accurate analyses for chloride electrometrically at the 10—20)ul sample range. Gasometric techniques are 42 available for determination of carbon dioxide on 10}ul samples though these are time consuming and comparatively clumsy. Gas—chromatographic systems are available also but impractical for most laboratories. Instrumentation currently exists for determination of pH, pC02,and p02 on a total of approximately lOAJl whole blood. A calculated carbon dioxide value could be derived from the first two measurements. Recent developments in methodologies have permitted the analysis of many constituents without prior protein separation. This avoids sample loss and difficulties with filtration, or a further separation stage following centrifugation. Special techniques must be employed to obtain a meaningful creatinine measurement, to eliminate interference from non— creatinine chromogens. The sample volume required for an accurate creatinine determination is still greater than 25 pl. Most other proce- dures have been adapted to sample volumes smaller than this. 2. Currently Used Equipment Analytical "systems” are commercially available, based on the original concepts of Sanz5 for analysis of many different constituents in small volumes of serum. Beckman/Spinco market a range of such equipment including pipettes, containers, centrifuge,and spectrophotometer which permit analyses that are reproducible by conventional manual techniques. Eppendorf has equipment available for this purpose including pipettes of a different design and a specially designed spectrophotometer. In both systems, emphasis has been placed on accuracy of pipetting and colorimetry, which are the most critical events in performing an accurate measurement on a small sample volume. 5M. S. Sanz, "Ultramicro—methods and Standardization of Equipment.” Clin. Chem. 4:406-419, 1957. A3 The Autoanalyzer has been used for making analyses on small sample volumes, with a precision claimed to be similar to that using the conventional sample volumes. The required precision of pipetting of sample into the system, however, is unlikely to be achieved without modifications of the pump and sampler units unless pre—dilution of the sample is used. Timed air—segmentation and miniaturization of components, as used in the SMA 12/60 and now available in the recently announced SMAR/micro system,should enable precise meaSurements to be made for many tests simultaneously. 3. New Eguipment Many other systems are at various stages of development or have been introduced to the market recently, and their performance has not yet been fully evaluated. The fast centrifical clinical analyzer GeMSAEC, originally developed by scientists at Oak Ridge National Laboratories under an interagency agreement with the National Institute of General Medical Sciences, is now under commercial development by at least three companies in the United States and one in Great Britain. Although not fully evaluated, this instrument if successful should be able to perform large numbers of tests using micro quantities of blood. Many of the other systems are of foreign manufacture and are not widely distributed in the United States, e.g., Baird and Tatlock Analmatic Clinical Analysis System, Griffin and George Bioanalyst, Norelco/Unicam Instruments AC System, and Joyce-Loebl Mecolab. The Grant-Linson Autolab, and Quickfit 617 Automatic Analyzer, available in the United Kingdom are capable of processing samples as small as,to smaller than,25}ul of serum throughout different stages to a colorimeter with presentation of data in several different 44 forms. The Quickfit6l7 can be used with whole blood. A centrifuge which is automatically loaded and unloaded is included in the system. The Analmatic system includes a separate centrifuge. Both of these systems permit protein separation, but only the Analmatic is capable of simultaneous analysis of the test and blank solutions. Micromedics in the United States and Hitachi, Japan,are currently developing discrete sampling analyzers. Scientific Industries (USA) is developing a novel system which adds serum to reagent-impregnated filter pads mounted on a continous belt. LKB (Sweden) is marketing components which may be interfaced to form systems for colorimetric analyses. The LKB Reaction Rate Analyzer has the capability of kinetic enzyme measurements on small samples. Carlo Erba Spa manufactures an automatic chemical analyzer, Series 1500,which is a continuous flow apparatus similar to the Technicon Autoanalyzer. Nearly all of these systems are in prototype form, or in the early production stage; and authentic information on their actual or potential use is not yet available. E. Emergency or Trauma Stations Present trends in automation make centralization of clinical laboratories attractive because they result in the most economical use of expensive equipment and skilled personnel. Such centralization does not, however, always meet the needs for rapid emergency service in treatment of critically ill patients. Valuable time is lost in trans— porting specimens from the patient to the laboratory if the central laboratories are located at some distance from the emergency room. The instrumentation of the trauma laboratory should be developed with the following general considerations in mind: 45 l. a minimum of specimen handling 2. simplicity of operation including automatic calibration 3. completely self-contained except utility connections including sufficient reagents for a specified period of operation 4. utility requirements limited to electrical, tap water and waste 5. capacity to record and enter data into data-processing channels for billing, patient summary, and physician identifibation 6. compact design to permit maximum mobility 7. modular components easily serviced and repaired 8. maximum reliability of instrument Operation with automatic alert for malfunction The chief aim would be to design a laboratory where technical personnel could provide the physician with reliable emergency determinations in the shortest possible time, preferably within two minutes. Test selection in the emergency situation should be limited to determinations required for immediate patient care. Compact design of the equipment and minimum utility requirements would permit Such equipment to be relatively mobile for movement into areas where intermittent need for rapid service might arise (i.e., intensive care unit, coronary care unit, surgery, etc.) Such instrumentation would ideally serve the small hospital laboratory where emergency service is a major function. Biochemical monitoring of vital parameters in intensive care units has been accepted for many years. However, surgical trauma is associated with many different biochemical responses, the extent of which is often over- looked postoperatively and may be responsible for "biochemical deaths.” 46 The ideal analytical system would not require withdrawal of blood an¢ indeed, sensors have been developed for monitoring the circulating P02 and PC02 through the skin. An indwelling catheter for periodic sampling of blood, coupled to appropriate analytical instruments and to a computer, should make possible a feedback loop to control infusion of the appropriate electrolytes and to maintain optimum electrolyte and acid base balance in the traumatized patient. Techniques should be explored for using saliva and expired gases to avoid even the need for indwelling catheters. A7 CHEMICAL TOXICOLOGY For practical purposes, the chemical toxicology laboratory is concerned only with laboratory specimens from patients with disability or potential disability due to an exogenous chemical agent. In a hospital these exogeneous chemical agents are usually drug substances that predominantly affect the central nervous system; heavy metals, pesticides, rodenticides, carbon monoxide, and other noxious gases as well as a variety of substances in non-drug commerical products to be found in or about the home. The laboratory effort applied to such specimens includes, in order of importance and priority, identification of the exogenous agent, quantification of the agent, and if requested, interpretation of the significance of the qualitative and quantitative resultsobtained along with available information about specific or nonspecific treatment of the disability. Many of the procedural aSpects of clinical toxicology are directly applicable to problems encountered in all areas of legal and industrial toxicology. Interpretive Aspects A central data bank of toxicological information should be established to contain all the analytical characteristics of different drugs. State and Regional Toxicological Laboratories should be equipped with telephone terminals to obtain rapid identification of toxicological agents. This would encourage uniformity of analytical procedures that would ultimately upgrade the quality of the work. For maximum helpfulness to physicians, the data bank should also store all the characteristics of known drugs, including appearance of the substance, signs, symptoms, and body fluid concentrations in over-dosed patients and indicate immediate and long-term therapy. 48 Analytical Aspects Until 1950, toxicological analyses were essentially unavailable as routine tests. In a few medical centers, there were individuals who performed occasional analyses of poisons for hospitals or private physicians, or more frequently, for medical examiners; but the scope of such services in diagnosis, and care of patients, was sharply limited by lack of sensitivity and/or specificity of methods. In the case of drugs, most analyses were limited to fatal cases, or to stomach contents obtained soon after non-fatal ingestions. The progressive increase in the number of drug products after the Second World War and their use and abuse was attended by improvements in photometric instrumentation that greatly increased sensitivity in many drug analyses, for example, UV photometry, and specificity of identification, for example, infrared photometry. At the same time, separation techniques involving the various methods of non-gas chromatography were effectively exploited as were other approaches, such as counter—current distribution, paper electrophoresis, Conway diffusion, etc. By 1960, the application of gas chromatography to analytical toxicology had increased the capability for toxicological analyses. It was still sparsely available, however, in hospital settings for at least three reasons: 1) The instrumentation required was disproportionately costly for the amount of patient service delivered; 2) Methodologies requiring complex preparation and instrumental measurements were infrequently called for, and Such methods were difficult to keep operative; and 49 3) Very few individuals were trained in toxicological methods. Within recent years, opportunities for exposure to toxic Substances have grown at an alarming rate. The use of drugs for non-therapeutic purposes, of pesticides, chemical food additivies, and alcoholic beverages has increased. There are now about half a million commercial products containing potentially toxic ingredients that are used in industry and about the home. Until recently, the response to increasing demands for toxicological analyses by clinical chemistry laboratories has been slight. Although a few large hospital laboratories in metropolitan areas provide extensive services, most others are equipped to perform only a limited variety of tests or none at all. Requests for service determinations must be referred, usually to a medical examiner or to a chemist trained in clinical toxicology, and there are very few of these well-trained specialists available. Future Developments There is no doubt that mechanization of toxicological analyses in large hospitals or regional laboratories would improve service materially in terms of number, accuracy, and rapid performance and regular availa— bility of tests. A major decision must be made as to whether toxicological services are best provided by a few large regional laboratories equipped and staffed to perform a broad spectrum of analyses or whether they should be available in every hOSpital. It seems likely that the former arrangement will be chosen. With modern shipping and communication capability, no hOSpital would be so distant from well-spaced centers 50 that the time required of specimen transfer would be a critical limitation. The alternative of developing adequate capability in each hospital or each community is unlikely because of the cost per unit of service imposed by equipment and personnel requirements. The best compromise for the present would be to encourage individual laboratories to develop toxicological capabilities to satisfy their most pressing needs. Determinations not offered by a laboratory could be referred to the nearest facility with competent staff. The choice of tests to be performed in a limited operation is difficult but is facilitated by the available experience and statistics on disabilities caused by exogenous chemical agents. For example, figures from the Department of Health, Education, and Welfare show that approximately two— thirds of fatal poisonings are due to three agents: barbiturates, carbon monoxide, and ethyl alcohol. Knowledgeable clinical toxicologists believe that the following substances or classes of Substances account for about 90 percent of exogenous chemical agent disabilities in patients admitted to hospitals: Barbiturates Carbon monoxide Non-barbiturate sedatives Heavy metals Salicylates Cholinesterase inhibitors Tranquilizers Kerosene, naptha Opium alkaloids Amphetamine Aliphatic alcohols Mechanization or automation of two kinds of toxicological analyses should be encouraged and supported. (1) Unit or class—limited instruments, largely for performing specified analyses or operations such as mechanized determination of alcohols and some other volatiles, high pressure primary separa- tions, solvent extractions,and concentrations. 51 A number of these are already operational, e.g., the Perkin-Elmer mechanized head-space gas chromatographic analyzer for determination of alcohols in blood developed from the prototype designed by Machata6, the Dupont ACA for determination of salicylate and the Autoanalyzer for determination of ethanol and selicylates in blood. Several other devices, in addition to those already mentioned, include pressurized gel filtration, ion exchange or other chromatographic separations. Priority of development should be given to instruments to deal with the most frequently offending poisonous agent. The development of simple kits to measure the common poisons should be given a high priority. A minimum amount of time should be required for sample preparation and reSults must be obtained rapidly. The methods employed in reagent kits should be capable of determining the class of the compound but not necessarily its actual identity. Such kits should be available in all hOSpitals removed from the large centers with extensive toxicological facilities. The capability of class limited instruments would be enhanced by mechanization of a sequence of the operational steps common for each class of compounds. For most acidic, basic, and neutral drugs with low vapor pressures, the sequence could involve measurement of the liquid sample (e.g., blood or serum), primary separation of low molecular weight compounds (e.g., by gel filtration), organic solvent extraction at programmable pH's, solvent concentration, programmable treatment of 6G. Machata, "Uber Die Gaschromatographische Blutalkoholbestimmung." Blutalkohol, 4:252-260, 1967. 52 aliquots of the solvent concentrate with derivatizing agents and additions of internal standards, and transfer of a meaSured aliquot of the treated solvent to columns of a gas chromatograph. It appears that at present there are no satisfactory designs for achieving the final step of a sequence. For heavy metal analyses, there are good prOSpects for a mechanized sequence of sample measurement, addition of a chelating agent, solvent extraction and concentration, and atomic absorption photometry of the concentrate with a pre—set lamp sequence. It would seem feasible to adapt the Machata analyzer for alcohol to the detection and quantitation of volatile compounds generally including carbon monoxide. (2) Broad spectrum analyzers combining acquisition of data with electronic searching of data output and identification and quantitation read-out: Parts of mechanization sequences are seen in several recently announced instruments. For example, in one gas chromatograph system, samples are automatically processed and injected although they are not prepared for injection. Complete mechanization of such instruments is a formidable and inherently expensive task. It seems likely instead that improvements will come from the mechanization of techniques now used for the detection of low~vapor pressure acidic, basic, and neutral drugs. Several gas chromatographic-mass spectrometer interfaces have already been described. The sensitivity achievable with this combination of instruments has already been enhanced by development of a diffusion separational device which removes much of the carrier gas and enriches the component concentration in the fraction entering the spectrometer. 53 Alternatively, gas chromatographic column fractions may be directed into other interfaces. An already described infrared interface permitting continual rapid scanning of peak components would appear more effective than ordinary infrared photometry following deposition on a surface, or long light path infrared photometry. Direction into polar or non-polar solvents for high resolution UV scanning would not appear to have the potential usefulness of other interfaces. It seems unlikely that inter- facing with a spectrofluorometer would be preferable to infrared analyzers or mass Spectrometers. The use of fluorometry would probably be limited to materials not amenable to gas chromatography, with mechanized equipment carrying out primary and secondary separations prior to the interface with the fluorometer. Electronic data processing of various instrumental outputs has already been described, for example, by Kazyak7 for UV and IR data, and 8 for the more complex problem of translating mass Crawford and Morrison spectrometer data into a defined structure. Even aSSuming the development of the most sophisticated broadspectrum analyzers, it is unlikely that any can be flexible enough to do everything. A variety of unit or class-limited instruments will be required for maximal service. 7L. Kazyak, "Laboratory Automation in Analytical Toxicology." let Ann. Meeting Acad. Forensic Sci. Chicago, Ill., 1969. SL. R. Crawford and J. D. Morrison, ”Computer Methods in Analytical Mass Spectrometry, Structure Codes in Processing Mass Spectral Data. Clin. Chem. 41:994—998, 1969. 54 HEMATOLOGY General Aspects Hematology is a division of laboratory medicine limited to a single physiological system -- the blood forming and transporting system. All laboratory disciplines are utilized. In this respect, hematology is a medical sub-specialty, differing from the technical sub—specialties of clinical chemistry and microbiology,each of which applies a Specific laboratory technology to physiological systems and medical problems. Hematology,although relatively limited in medical scope,utilizes a broad armamentarium of methods. This sub-specialty is also more clinically oriented than the other laboratory sub-specialties and involves personal correlation of results, interpretation,and clinical involvement by the laboratory scientist. Laboratory techniques for hematology are fewer in number,and instrumentation is not as varied as in the clinical chemistry laboratory. Many hematology methods are rather well standardized. Mechanization or semi-automation has progressed so that 50 to 60 percent of the workload in a large laboratory can be handled efficiently and rapidly with increased accuracy by commercially available computer assisted programs. Instrumentation Instrumentation for hematology includes five classes of devices: cell or particle counting, direct and indirect; coagulation end-point detection methods; colorimetry; radioactivity detection; counting instruments and pattern recognition methods. Each class of instrumentation will be reviewed and the related needs for further development will be emphasized. 55 Cell and Particle Counting Devices: The electronic direct particle counting instruments designed and manufactured by the Coulter Company provide the most accurate and reproducible results for blood cell and platelet enumeration. Both number and size (volume) of cells can be measured, and the current Model S provides seven direct or derived measurements in one sample assay operation. Size distribution by use of a pulse-height analyzer is available. Other instruments (Technicon SAM 4 and 7, Fisher Autocytometer, etc.), utilize direct counting principles,such as optical detection or indirect electrical field effects for estimating derived numbers. Inherently, indirect derivations of cell counts tend to be less accurate and reliable; and the continuous flow multiple measurement instruments have, so far, produced consistent deviations for cell volume measurement. Several of these multitest counting instruments include measurement of hemoglobin concentration as well. Several attempts have been initiated recently, but without Success, to develop a reticulocyte identification and counting machine. The instrument under development at the National Institutes of Health,Clinical Center,Pathology Department appears promising if sufficient development can be financed to solve several optical and mechanical problems. In cases of anemia, the absolute reticulocyte count is a useful measure of the capability of the bone marrow to reSpond to therapy. Following the reticulocyte count after the initiation of specific therapy (iron, folic acid, or vitamin B12), is the best way to evaluate efficacy of the therapy. The reticulocyte depression is a warning index of too much drug therapy, in the chemotherapy of leukemias or in the use of chloromycetin, for 56 example. A high reticulocyte count without anemia may indicate a "hemolytic state" which should then be purSued with more specific diagnostic procedures. Devices for Detection of Coagulation: Several electro-mechanical instruments are commercially available for timing coagulation of blood plasma and various mixtures of coagulation factors, prothrombin time, etc. Being a series of enzyme—like kinetic reactions, the exact coagulation end—point is necessarily arbitrary in definition and difficult to measure accurately or standardize. Of the several devices tested by five leading hematologists, the most reliable and acceptably "accurate” was the Fibrometer (BioQuest). This instrument measures the time interval during incubation from mixture of the desired reagents to the moment when ”initial” fibrin thread formation is sufficient to form an unbroken bridge between the mixture surface and an electrode that dips into the surface repeatedly at a fixed rate. The fibrin thread conducts electric current which stops the clock via relay. This is a digital device which may be coupled to a computer, but sample identification is manual and visual and poses a problem for automation of this procedure. Accession number input keyboards can be connected to the instrument, but input of each sample identification number requires the continuing attention of the technologist. Only one instrument, the Fibrometer Lysis Sensor, is available for measurement of fibrinolysis, but so far, this has not proven satisfactory. These coagulation end-point meaSuring instruments are not truly mechanized. They require manual manipulation for adding necessary 57 reagents at scheduled and sequential intervals to provide required incubation periods before clotting starts. This is a fatiguing, error- prone operation and effective automation is needed. Future development should explore an entirely new approach. Coagulation end-point determination is empirical and grew out of manual manipulations and man's ability to viSually observe and judge the nature of the first evidence of clot formation. Coagulation and the interaction of the several coagulation "factors” however, appear to be a series of enzyme reactions. Determination of the properties of these reactions in health and disease by measurement and description of their kinetics may be more meaningful from the medical standpoint. With the advantages of computer processing and the more sophisticated enzyme kinetic measuring techniques which are available today, study of coagulation kinetics should be feasible and deserves emphasis, exploration,and developmental support. Another new approach deserving of exploration is radio-immune technology for detection of clotting factors. Colorimetric Methods: In this category, the value of hemoglobin concentration is most frequently requested. The technique is mechanized in several ways;and in a few instances, the output of the instrument is directly coupled to a computer. The Coulter Model S and SMA 4 include hemoglobin colorimetry. For several enzymes (G-6-PD, phosphatases), serum iron, etc., manual colorimetric methods should be mechanized, particularly, G-6-PD determinations which would be made more frequently if mechanized. Serum iron and iron binding capacity have been mechanized using continuous flow instruments. Mechanization is valuable because it 58 enables the laboratory to perform easily the large numbers of determinations required to differentiate between the more frequent iron deficiency anemias and those of Thalassemia syndromes and microcytic hypochromia -- due to non—utilization of iron rather than to the absence of iron stores. Hemoglobin electrophoresis might be of value in studying certain population groups and can be at least semi-automated with instrumentation recently developed and now in various stages of evaluation. Another possibility is the electrophoresis of whole red cells as well as hemoglobin. If practical, one could expect to detect many of the hemoglobinopathies now overlooked, especially those involving hemoglobins S, C, and A2- Determination of the serum haptoglobins has been mechanized and is worth— while as an index of the presence of a hemolytic state or as a baseline value for reference in case of a suspected hemolytic episode. Radioactive Isotope Procedures: Erythrocyte survival assays, Schilling Tests for vitamin B-12 absorption, and blood volume meaSurements may be performed using radioactive methods. Radioactivity detectors and counters are well developed,and many are sufficiently mechanized to permit easy coupling to computers for raw data processing. Sample handling,at present is manual,and identification must be entered into the computer via type- writer or keyboard. Except in centralized or regional laboratories which handle a large volume of such tests, the expense of automated equipment does not appear to be justified at this time since the smaller laboratories are asked to perform only a few Such tests each day. Serum vitamin B—12 analyzed by the radioactive technique can be automated, but unless equipment could be produced at low cost, it is unlikely that many laboratories w0uld be able to afford this approach. 59 Most macrocytic anemias are due to folic acid deficiencies and diagnosis is usually made by a combination of history, exclusion of pernicious anemia, and a clinical trial of minimal doses of folic acid with Subsequent attention to the reticulocyte response. Cell Morphology and Pattern Recognition Devices: Blood cell classification is one of the large volume demands of the hematology laboratory. Cell identification related to function and number is one of the most important universally demanded test procedures. Differentiation of cell types in the peripheral blood and bone marrow is based on morphological patterns of cells and their structures as revealed by traditional staining methods and microscopic study. Recognition and classification is a high speed intellectual function for trained staff but a very slow and tedious procedure for even a sophisticated computer with an appropriate pattern recognition program. At the present time, there are a few mechanized systems available for differential blood cell counting and bone marrow examinations;but these are in early developmental stages. To date, machines which attempt to differentiate histochemically stained cells have not proved successful. In spite of great effort applied to design and development of computer systems for cell pattern recognition, the limited technical successes have been far too costly and slow for practical use. Future development must emphasize exploration of new approaches to cell classification based on functional and chemical characteristics instead of, or in addition to, selected morphological measurements or pattern parameters. Taking advantage of high speed computer processing for correlation of numerous data, it should be feasible to measure 60, automatically chemical and/or biological functional (enzyme kinetics, antibody—antigen reactions) properties and reactions simultaneously in rapid sequence. The resulting data could be processed in milli-seconds to classify cell types which could be related to physiological and pathological states. Such a system will involve many difficult problems, such as separation and sorting of cell types. While methods for cell separation are crude, there has been some progress in sorting and concentrating different cell types. A mechanized blood smear staining machine (Ames, Hematek) has substantially improved the reproductibility and standardization of staining Smears. Smear identification, loading and unloading are manual, but the sequential and mechanized staining of a series of 25 or more slides has reduced labor and staining time. Automated methods for the measurement of osmotic fragility would be useful as a screening procedure in communities of high risk for Thalassemia or the hemoglobinopathies, but the yield of abnormal determinations would be extremely low in most population groups. Data Processing There are a number of card systems for sorting and storing hematological data. An excellent comprehensive punchcard system was developed at the National Institutes of Health using mark-sense Hollerith cards. This provides an efficient, economic procedure for test re5ult input, storage, reporting of printed test reports, laboratory lists, worksheets,and summaries. Commonly available IBM electric accounting machines, such as key punches, sorters, and printers are used and Such a system is appropriate for Smaller hospitals and clinical laboratories. 61 Computer assisted systems are more expensive, and those commercially available vary in practicality for routine laboratory needs. In most instances, the hematology data processing has been linked to a system primarily designed to handle chemistry laboratory needs. This usually consists of indirect coupling of a hemoglobinometer, Coulter Counter, or Technicon SMA 4 to a small computer via punchcard or punched paper tape. There are a few systems which include direct coupling by hardware interfaces. Hardware interfaces for direct, on-line coupling of the several hematological instruments to the computers have been developed for these two systems. Parallel to the electronic interface equipment are incremental magnetic tapes for temporary storage of raw data as backup in case of failure of the processing system. These interfaces buffer, coordinate,and transmit blocks of coded data including identification of the sample, the test type, and the raw result. Quality assurance programs are rather elementary, because sophisticated and comprehensive control requires complex feedback systems of hardware and software. Most quality assurance programs consist of printed lists of results on patients' samples and control sera representing distribution and identification of normal and abnormal results. Statistical analyses usually are limited to computation of coefficients of variation and histograms of weekly or monthly accumulated data. Medical interpretation programs for hematological data are limited to comparison of current test results to previous recent reSults of the same test on the same patient for evaluation of the significance of changes, and Such programs are provided by very few vendors. 62 Future development of hematological data processing will parallel improvement and innovation in instrumentation, methodology, and computer applications. There is a need for more sophisticated and extensive programs for correlation and interpretation of data to alert physicians to unusual results and to contribute information of value to diagnostic and therapeutic decisions. 63 BLOOD BANK General Aspects Several computerized blood bank systems have been developed and are in operation in a few large blood banks. Computers are used for blood inventory, outdating, use inventory, prediction, and donor sorting and searching. These are fairly satisfactory,and it can be expected that, as experience accumulates, their users will make needed improvements. Still urgently needed, however, are reliable, accurate and reproducible automated methods and instrumentation for blood cell classification and compatibility tests. The large volume of work in almost all blood banks justified considerable expenditure of effort and funds to Support this development. Immunohematological Methods Erythrocyte agglutination and hemolysis techniques are used widely and in large volume for determining cell compatibility for blood transfusions, platelet,and leukocyte rich plasma infusions and tiSSue or organ transplants. For cells and tissues other than erythrocytes, however, both methods and interpretation of reSults are uncertaingand the modest demand for these tests to date does not justify the costs of development of automation. Future development of transplantation techniques and benefits undoubtedly will increase demand and justify automation of the more reliable methods. Red cell compatibility tests are universally used in large numbers, and accurate reSults are critical to survival and health of patients. Most laboratories depend on well trained, dedicated and conscientious technical staff and carefully performed manual methods; but errors are 64 still frequent. Therefore, highly reliable and accurate methods and instrumentation for detecting and measuring antigen-antibody reactions are needed. In this field, positive specimen identification is especially important. The several attempts by industry to mechanize the ABO and Rh blood group tests have not proved reliable for routine work. This problem urgently needs adequate developmental Support. ea MICROBIOLOGY General Aspects The development of mechanized procedures for microbiological techniques has lagged far behind clinical chemistry. The reason is that most of the definitive tests in microbiology depend upon observations of the morphology, growth characteristics, and metabolism as well as the biologic interactions of a great variety of microbial species. None of these traditional procedures can be mechanized readily. Because microorganisms grow slowly on artificial media and must be isolated and identified to provide data useful for diagnosis and treatment, there is very little potential for mechanization in microbiology. The growth time in primary cultures and sub—cultures is so long that mechanization of manual handling manipulations would do little to shorten the turn-around time for clinically useful information. Judgement of the medically experienced microbiologist is required to sort out the clinically significant organisms from among the many non-pathogenic organisms in the mixture usually found in clutures of exudates, throats, stools, etc. Specific identification of pathogenic organisms is a tedious process. Moreover, even when definitive chemical reactions occur, they are complex and often represent only a small part of a continuum of a Spectrum of biological and chemical events. Clinical microbiologists have been properly more interested in the biology of the pathogens than in automation and are not by nature or training concerned with machines, mathematics, engineering principles, or data processing. Research in bacteriology and immunochemistry usually does not require automation and has thus provided no significant stimulus 66 for mechanization. Furthermore, during the era of ”miracle drugs," the antibiotics were mistakenly believed to have solved the problems of therapy so interest in medical microbiology was surpassed by interest in the other clinical laboratory sciences. Time has shown, however, that this appraisal of the value of antibiotics was overly optimistic. The increasing frequency of antibiotic-resistant strains of bacteria and the consequent demand for antibiotic sensitivity testing has revived interest in this discipline. Clinical Requirements and Feasibility Vastly improved services are needed in the microbiology laboratory. The clinical urgency for introducing definitive therapy early in an infection and for avoiding later antibiotic resistance requires improved methods for early detection and Specific identification of pathogens and for determining their sensitivities to specific antibiotics. There are a number of problem areas demanding priority development if the laboratory is to meet the most frequent and serious needs for patient care. Rapid isolation and identification deserve intensive efforts. The use of growth stimulators and ultramicro detection systems employing fluorescent antibody techniques for identification seem most promising. With computer accumulation of large amounts of specific test data and correlation with local community epidemiological data, appropriate programs could generate specific therapeutic guides for each case and avoid the long delays while waiting for bacterial growth and isolation from Subcultures. Further automation of the MacLowry9 microtiter method for determining 9J. D. MacLowry, M. J. Jacqua, and S. T. Selepak, ”Detailed Methodology and Implementation of a Semi—Automated Serial Dilution Microtechnique for Antimicrobial Susceptibility Test." J. Appl. Micro. 20:46—53, 1970. ‘ 67 antibiotic levels in blood and other body fluids would greatly facilitate rational treatment by daily monitoring of the concentration levels in blood and accumulating the response data in computer records to be used as guides in later cases. Serology This group of tests should be amenable to mechanization since quantitative pipetting and optical reading of agglutination reactions are mechanically feasible by modern technology. The volume of tests for the detection of syphillis, for the rheumatoid factor for coagulation, for agglutination titers against viruses, and for blood compatibility tests in the blood bank more than justifies an intense effort toward mechanization of these procedures. Fetal Antigens and Cancer In many cancers in man and in laboratory animals, new antigens appear which are not found in normal tissues in animals of comparable age. Some of these are auto—transplantation antigens which may trigger an immune response. Each of them in principle could be made the basis of a routine test for specific type or class of cancers. Examples are the alpha 1 fetal protein described by Abelev10 which occurs in serum of a large fraction of primary liver carcinomas, and the Gold11 antigen which is seen in serum of most patients with cancer of the colon. These antigens are found in the normal embryo or fetus, but disappear before or shortly after birth. With the successful removal of a tumor of the colon, the 10G. I. Abelev, "Antigenic Structure of Chemically—Induced Hepatomas.' Prog. Exp. Tu. Res. 7:104—157, 1965. 11P. Gold and S. 0. Freedman, ”Specific Carcinoembryonic Antigens of the Human Digestive System." J. Exp. Med. 122:467—481, 1965. 68 circulating antigen disappears whereas incomplete extirpation is signalled by continued presence of the antigen. Recent research suggests that a large number of different embryonic antigens are re-expressed in cancer cells. We foresee a very large requirement for assays for embryonic antigens for early cancer detection, and for the evaluation of therapy. The techniques are often difficult and complex, and should be automated. Current Progress in Mechanization Although Wilson and Junger12 list certain procedures for bacteriological cultures and colony counting as "fully automated and available for high speed,” no published description of such methods can be located. They also list partially automated tests for the rheumatoid factor, VDRL, and antistreptolysin. Bacteriological Cultures: Sophisticated, mechanized large volume culture methods and computer assisted colony identification and counting based on optical methods have been developed by Glaser13 at the University of California, Berkeley, for genetic studies with support from the National Institute of General Medical Sciences. Spinsoff from this excellent work may contribute to automation techniques for clinical application. To date, this approach has proven capable of identifying pathogens with considerable accuracy in pure cultures, and similar applications in mixed cultures appear feasible. Some attempts have been made to mechanize the cultivation of bacteria from urine. So far, these attempts have not been particularly lzJ. H. G. Wilson and G. Junger, "Principles and Practice of Screening for Disease." WHO, Geneva. 34:141, 1968. 13D. A. Glaser and W. H. Wattenburg, ”An Automated System for the Growth and Analysis of Large Numbers of Bacterial Colonies Using an Environmental Chamber and a Computer—Controlled Flying Spot Scanner." Annals of the New York Academy of Sciences. 1392243—257, 1966. 69 14 successful. Sharp and Keen emphasize the importance of early detection of urinary tract infection, but mention only tetrazolium and nitrite tests performed manually to speed recognition. These authors comment on the lack of reliability of these methods. They also describe a simple, rapid, manual method for plating and counting colonies as a semi-quantitative method. Bowman, Blume, and Vurek15 have developed a micro method for rapid culture of organisms and quantitative estimation of bacteria in biological fluids by an automated method. Small capillaries are mechanically filled with a culture medium mixed with the sample to be tested and then incubated and periodically monitored by a light scattering detection device. Micro colony counts are available in a few hours. Practical applications of this approach are now being evaluated for estimating numbers of bacteria in liquid culturesl6. Both densitometers and electronic particle counters, have been tried with questionable success. Necessary nutrient materials in the culture media produce interfering particles and prior washing of the bacteria causes clumping. Hochberg and Caceresl7 mentioned that grouping of bacteria, and presumably their identification by automated fluroescent methods, are feasible and have shown great promise. 14 . 'C. L. E. R. Sharp, Presymptomatic Detection and Early Diagnosis: A Critical Appraisal, C. L. E. H. Sharp and Harry Keen, Pitman Medical, London, 1968. 15 R. L. Bowman, P. Blume, and G. G. Vurek, "Capillary Tube Scanner for Mechanized Microbiology." Science. 158:78, 1967. 16 F. R. DeLand and H. N. Wagner, "Early Detection of Bacterial Growth Using Carbon-l4 Labeled Glucose." Radiology 92:154—159, 1969. 17 H. M. Hochberg, J. K. Cooper, J. J. Redys, and C. A. Caceres, "A System for Machine Identification of Bacteria.” Medical Research Engineering 7:20—24, No. 2, 3rd quarter, 1968. Identification: Some early work by Gunnar Junger on the automated measurement of bacterial growth products correlated by computer data reduction as a means of identifying certain pathogenic bacteria seemed promising, but the work has not been pursued (unpublished, 1960). Several recent attempts to apply gas chromatography to the measurement if combined with other measurements such as end—space gas analysis, but the difficulties of separating components from products and culture media contaminants and the similarity of results for different strains has prevented useful progress. If improved, this technique could lend itself to automation, but would still require serial handling of samples. Antibiotic Sensitivity: By contract with Aerojet-General, the Aerospace School of Medicine, USAF, has developed a breadboard model of a machine for antibiotic sensitivity testing of isolated strains of organisms. The mechanism is capable of handling 100 culture tubes in series and transporting them through an incubator for programmed periodic readings of turbidity by a densitometer. Growth curves are identified and compared by computer for varying concentrations of antibiotic. They have demonstrated satisfactory correlation of 24 hour minimum inhibitory concentrations with data from the first eight hours of growth. Pipetting of cultures and antibiotics is manual, however, and organisms must be cultured, subcultured and isolated before testing so there is no saving of time. The machine is useful for high volume testing of a few organisms. - , 18 At the National Institutes of Health, MacLowry has developed and is 18 Methodology and Implementation of a Semi-Automated Serial Dilution Microtechnique for Antimicrobial Susceptibility Test." J. Appl. Micro. 20:46—53, 1970. J. D. MacLowry, M. J. Jacqua, and S. T. Selepak, "Detailed 71 successfully operating in daily patient care, a mechanized system for microtiter measurements of antibiotic sensitivity using tube dilution techniques. The microtiter equipment has been mechanized and is amenable to further automation. The manual process has been speeded up by a factor of 20 or more and the addition of reagents and culture media mechanized. Isolated cultures are still required. This method enables two technologists to perform 400—600 sensitivity tests per day easily in contrast to the 30 or less before mechanization. Previously,the tédious work of testing antibiotic levels in the blood with any useful degree of quantification restricted this service to one or two patients per day. Physicians now are encouraged to submit daily samples of blood in order to monitor the level of administered antibiotic and thus guide further therapy. This is the one Successful and real contribution of automation in the microbiological laboratory to date. A new machine for antigen—antibody reaction testing developed by industry is the Spectra Auto-I of Spectra Biologicals, Becton-Dickinson, for A, B, O, and Rh testing of blood. The principles are sound but there are some potential mechanical and measurement difficulties in design. With sufficient bench testing and debugging, this machine may prove useful in microbiology as well as immunohematology. Phage typing has been partially mechanized by Zierdt19 of the National Institutes of Health. This instrument permits simultaneous spotting of 24 different phages in measured amounts sequentially onto 190. H. Zierdt, F. Fox, and G. F. Norris, "Multiple Syringe Bacteriophage Applicator.” Am. J. Clin. Path. 33, No. 3: 233-237, 1970. 72 a series of plates containing various organisms. This has Speeded the phage typing operation by a factor of 15 or more, primarily by eliminating all mouth pipetting. The instrument is commercially available. Need for further automation is doubtful because phage typing is not an immediate patient care procedure. Data processing for clinical bacteriology has been partially accomplished in a few laboratories, but all represent reduction of data manually entered and generated by traditional manual methods in the laboratory. Reports are printed along with chemistry and hematology results. Parasitology and Mycology: No information is available to indicate that any efforts have been made to mechanize or automate methods in these areas . 73 VIROLOGY General Aspects One of the difficulties in dealing with virus disease is that currently the only procedures available for the isolation and identification of the etiologic agents are complicated, time conSuming, and expensive. Physicians, sometimes justly, complain that by the time the laboratory report reaches them the patient is either well or dead. Thus, the most important practical advances to be made within the next decade will be the development of rapid and inexpensive methods for recognizing the presence of Specific viral agents within tissues and body fluids. Many of the same problems confronting the mechanization and automation of procedures in microbiology apply here as well. Indeed, this situation may pose an even more difficult challenge since viruses thus far have not been cultivated in the absence of living cells. The number of different types of cells, however, that will support viral multiplication is ever increasing. These are available as a source of primary cells from whole organs of the human embryo and the kidneys of primates and rodents, to mention only a few examples, and from cells derived either from malignant or normal tiSSues which propagate them— selves ad-infinitum as cell lines. Future Developments A variety of instrumentsare being devised which will make possible the measurements of the number, size, and structure of microbial agents as well as the refractive indices of their component parts. The effects of physical and chemical agents on such particles would also be meaSurable 74 with a high degree of precision by these instruments. This should.be useful in the monitoring of microbial growth, anti-microbial susceptibility, phage growth in bacteria, and other such activities within the bacterial cells. For example, modification of light- scattering photometers may be introduced into instruments for measuring the patterns of immune reSponses, clumping, precipitation, turbidity, and the size and composition of the particles present in bacterial cultures. The computerized mathematical analyses of such patterns as compared to standard models would then be used to determine the specific identity of bacteria or their response to specific physical or chemical agents. Recently, with support from the National Institute of General Medical Sciences, a new method termed isopycnometry has been developed at the Oak Ridge National Laboratories which allows small numbers of virus particles and other antigens to be detected; their number estimated; and their mass measured. This method potentially will allow the rapid diagnosis of viral diseases. Intensive efforts are underway at this writing to adapt this method to routine use and to develop the instrumentation required. SUMMARY AND CONCLUSIONS Good health care requires good laboratory support. A large fraction of the benefits of medical research can only reach the patient through the clinical laboratory which has traditionally acted as a bridge between science and medicine. To provide the quality and quantity of laboratory services needed for patient care, clinical laboratories must aim at three goals: (1) Delivery of uniformly high quality laboratory service to all citizens; (2) Expansion of mechanized laboratory processing to provide quality service at acceptable cost; and (3) Correlation of laboratory and clinical information for rational and effective use. Corollary to these basic needs are requirements for standardization of methods, quality assurance and reporting, development of better and faster assay machines and methods that use innovative techniques, and recognition of early and Subtle deviations of chemical and physiological data portending disease. To develop a rational approach for the complete automation of clinical laboratories, a major effort on the part of concerned laboratory scientists is needed to achieve four general goals. I. Upgrade existing systems and approaches. This objective is dictated by continuing increase in workload, and is being approached through interim solutions because many laboratory tests now in use may ultimately be replaced. Until replacement tests can be devised and validated, laboratories must make do with whatever improvements 76 are now possible. Some of these involve mechanization or automation of present methods, better sample identification, miniaturization, and the use of computers. 11. Develop new methods, procedures, instrumentation and systems for bringing the clinical laboratory to the level of excellence and productivity that will meet national health requirements. The reservoir of scientific knowledge and technology available has not been systematically applied to the solution of these problems, but could be so applied to great advantage. Clinical laboratory scientists must be attentive to the public expectation that the findings of basic studies, largely gained at public expense, will be applied to the improvement of the public health. Their responsibilities include the development of reliable criteria for choosing those concepts and projects most likely to yield substantial improvements, the adOption of administrative techniques and procedures to implement selected programs, and the development of methods for objectively assessing final patient benefit. III. Identify and suppprt basic research and development in areas of high promise. While many marked improvements can and must be obtained from the direct application of present knowledge, it is evident that the new, and sometimes revolutionary developments of the future will be based on knowledge we now lack. The broad outlines of these advances can often be envisioned clearly enough to identify the nature of the basic research that must be done to achieve them. When this occurs, progress may be markedly accelerated by judicious and thoughtful choice and adequate funding of innovative projects. 77 IV. Standardize reference materials, methods, and units. Useful measurements require a frame of reference to assure validity. Accordingly, all branches of laboratory medicine must have ready access to well defined reference materials for calibration of equipment. Standardized reference methods must be developed for comparison of analytical results and for evaluation of new techniques. A uniform and comprehensive system of reporting laboratory information should be established so that: (A) all quantitative data are expressed in uniform numerical units; (B) the analytical technique used is so Specified that the data for each substance or parameter can be compared with that obtained by other analytical techniques; (C) the drift of the data from the state of health can be quantified statistically; and, (D) trends in changes for each test and clusters of tests can be interpreted objectively by using computer techniques. In the course of development of this study, the committee repeatedly was reminded of the important role that well trained clinical scientists occupy in assuring the proper delivery of laboratory data. There is no question but that maximum utilization of the clinical laboratories will depend on adequate staffing by well trained personnel at both professional and technical levels. The expanding variety of scientific knowledge applicable to clinical laboratory problems makes it unlikely that traditional academic procedures will produce the needed skills. New concepts for training laboratory Specialists must soon be formulated and deve10ped. 1., fin... ‘ -- i». V... .m—h ' I _ ‘ ”5.9“" .—---..<. ‘.. fl \ ...»\.._..--....- ~ . ..-....,___, _,_,_,“. Q‘f‘l' US. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service National Institutes of Health DHEW Publication No. (NIH) 72-145