Valid Information From Sputum

Don’t Throw Out the Baby With the Bath Water

The investigation by Rubin that appears in this issue of CHEST (see page 195) brings to our attention the potential value, as well as the difficulties, of studying the sputum expectorated by patients. Sputum provides pathologic material that is a direct reflection of the disease process that led to its production in the bronchopulmonary system. When produced, sputum is always abnormal, and it provides a recurring “biopsy” of the pathophysiologic process. The knowledge to be gained from sputum was well appreciated from the end of the 19th century through the 1920s and had been comprehensively compiled by von Hoesslin in 1926¹ and by Benzançon and DeJong in 1913.² More recent reviews have been edited by Dulfano in 1973³ and by Takishima and Shimura in 1994.⁴

Studies that have defined the various characteristics of sputum have examined its physical properties, clearance mechanisms, chemical and cellular compositions, and microbiology. However, the application of these studies to clinical practice has taken place mainly in the assessment of infection. The role of microbiological techniques in defining specific bronchopulmonary infections has been the basis for the evaluation of their medical and social significance and for developing specific treatments. Sputum cytology for the detection of bronchopulmonary malignancy also occupies a respected place in clinical practice. However, the other measurable sputum characteristics, particularly pertaining to the assessment of nonmalignant cytologic patterns, have not been generally utilized in clinical or sputum research. This omission has occurred despite observations for over a hundred years that the nature of the inflammatory cell response and the characteristics of cells exfoliated or torn from the bronchial epithelium could identify the type of pathology and often point to a specific diagnosis. At a more basic level, a very simple microscopic evaluation of expectorated material can clearly differentiate the material that is the product of the bronchopulmonary system from the material that originates in the upper airways.⁵ Using prescreened aliquots for microbiological and cytologic tests is most apt to provide useful information. Further testing of inadequate material is a waste of resources. For example, the Murray and Washington criteria⁶ have significantly improved the reliability of Gram’s stains, but have only marginally improved the reliability of cultures.

Over the last few decades, clinicians have become less familiar with the potential value of sputum examination, which often leads to an indiscriminate distrust of results of all kinds of sputum tests. This state of affairs is at least partially related to the insistence of clinical pathology laboratories that testing carried out by clinicians on wards, in clinics, or in offices is not subject to the same quality control as when trained technicians perform the testing. Clinicians in training no longer gain experience in evaluating clinical specimens and, consequently, often cannot assess the information provided by the laboratory. It is understandable why clinicians distrust the information obtained from the microbiological evaluations of sputum. For patients to be managed appropriately, clinicians should expect their supporting laboratories to use the best techniques to provide reliable information.

Utilizing sputum in research is even more problematic, particularly when selecting the patient population, defining the disease(s) to be studied, determining the method and timing of sputum collection, selecting the appropriate sputum aliquots, choosing an evaluation methodology, and interpreting the results. Protocol design should consider these various factors so that the methodology will provide data best suited to test the goals of the specific investigation. An accurate definition of the patient population is important for all clinical investigations of bronchopulmonary disease, not just those investigations related to sputum. For example, if the patient selection for a study of COPD is based solely on symptom history and pulmonary function, there may well be a variable and undefined mixture of patients included with chronic bronchitis, chronic bronchial asthma, or both. Because the pathophysiologic inflammatory natures of bronchitis and asthma are significantly different, the results of studies that ignore this difference can be confusing and misleading. This is particularly important in therapeutic trials. Other important considerations include how the sputum is collected: is the specimen freshly expectorated in the laboratory, is it limited to morning expectoration only, is it collected over a 24-h period, or is it an induced specimen?⁷ There is little comparative data to decide the approach that is best. Nevertheless, protocol design should consider criteria that will most likely reflect the underlying pathology of the disease under investigation. My experience has been that sputum collected over a 24-h period is more useful, because subjects vary as to when they produce their most representative samples. Sputum collection over a set time period provides the opportunity for quantitative measurements as well as the commonly used qualitative indices.⁸

A microscopic determination that the expectorated material is the product of the bronchopulmonary pathology remains essential for the selection of test aliquots. Aliquot selection based on gross visual criteria is unreliable.⁸ It may be necessary to examine multiple aliquots to determine the uniformity of the specimen. Methods that utilize homogenization, mucolysis, centrifugation, storage by freezing or cooling, or transportation should always be validated against fresh sputum. When homogenization is a necessary step, the true bronchopulmonary material should first be separated from the oropharyngeal secretions that are almost invariably contaminated. The numerous techniques that are used to measure the physical properties of sputum are generally more precise than the material being examined. Sputum is notoriously nonhomogenous, and its non-Newtonian nature compounds the difficulty in assessing the results. Using sophisticated methodology, such as that employed by Rubin, to measure physical and clearance properties does not obviate the consideration of other potential problems. There are few comparisons of the various methods used to measure physical properties. Chodosh et al⁹ found little correlation between the results obtained using five different methods. However, when using one method, the effects of guaifenesin on sputum when administered to chronic bronchitic subjects was demonstrated.¹⁰

Rubin and the other investigators who have stepped into the morass of sputum research are to be congratulated for their pioneer spirit. Although sputum is a difficult material to study, the rewards to medical science and clinical practice can be great. This readily available pathologic material can provide an important insight into many of the inflammatory diseases of the lung.