Spontaneous regressions of cancer, rare events but repeatedly recorded, have led generations of investigators to seek explanations for their occurrence and therapeutic maneuvers to increase their frequency. When a number of patients undergoing cancer regression in the late 1800's were found to have con-current bacterial infections, F. Fehleisen in Germany and William B. Coley in the United States, as well as a small group of other physicians, attemptedto induce infections in patients with far advanced cancer. Although antitumor responses were seen, some dramatic, it was difficult to infect most patients, and, when an infection did occur, there was no way to control its severity. Coley, therefore, turned in 1893to the use of killed bacte-ria, and the mixture of Streptococcus pyogenes and Serratia marcescens that he and others used to treat cancer came to be known as Coley's toxins. Coley's work was well known at the time, and in 1934 the American Medical Association stated that Coley's toxins were the only known systemic therapy for cancer. However, with advances in radiotherapy and, subsequently, chemotherapy, clinical interest in toxin therapy diminished, even becoming controversial incertain quarters. Coley's results would have been lost had not his daughter, Helen

Coley Nauts, collected and analyzed the records of her father and other physicians from this country and abroad (1). In contrast to the eclipse of clinical interest in toxin therapy, laboratory studies of microbial products as antitu-mor agents have had a long and uninterrupted history. A wide range of microor-ganisms have been examined, from bac-teria, yeast and other fungi to plasmodia and trypanosomes, but most attention has been focused on three groups of organisms: Gram-negative bacteria, mycobacteria such as Bacillus Calmette-Guerin (BCG), and corynebacteria such as Corynebacterium parvum. One of the most dramatic and reproducible phe-nomena in experimental tumor biology is the hemorrhagic necrosis ofcertain mouse tumors that can be seen shortly after the injection of filtrates from cul-tures of Gram-negative bacteria. Murray Shear and his colleagues at the National Cancer Institute identified the active principle as a polysaccharide (2), and subsequent work showed that this com-ponent, also known as endotoxin or bacterial pyrogen, is a lipopolysaccharide (LPS) and a major constituent of the cell wall of Gram-negative bacteria. Clinical applications of Shear's find-ings were limited because LPS was con-sidered to be too toxic in humans. BCG and C. parvum, however, were subject-ed to extensive tests in cancer patients, with generally disappointing results. The