How Does a Cell Caught by a Virus Warn Its Neighbor? Isaacs, Lindenmann and the Discovery of Interferon (1957)

A forerunner, an inflated hope and a measured limit

The first caution concerns a forerunner. In 1954, in a French-language journal (C R Séances Soc Biol 1954;148:1700-1702), Nagano and Kojima reported that rabbit tissue infected with vaccinia virus contained non-infectious components that blocked skin infection; the components did not sediment in an hour of high-speed centrifugation. In 1958 (C R Séances Soc Biol 1958;152:1627-1629) they wrote that a cell-free supernatant from ground infected rabbit skin or testis blocked vaccinia infection, and that this did not come from immunity or trace antibodies. Watanabe's 2004 letter says the two groups' work was independent and used very different experimental systems. The 1954 report preceded the 1957 paper; the name, however, became established through the Isaacs and Lindenmann paper. I did not read a source that settles priority, so I call neither one "first."

The second caution concerns hope. Clinical testing of interferon in cancer began in 1971, when Hans Strander at Karolinska Hospital started studies with leukocyte interferon prepared by Kari Cantell in Helsinki; the first disease was osteosarcoma (Gresser's 1989 review). The early studies relied on historical controls; Gresser defends them, arguing that randomization was not considered feasible at the time, but historical controls limit how far the results can be interpreted. Supply was also narrow: according to a 1980 Time article, Cantell separated white cells from blood collected by the Finnish Red Cross and stimulated them with Sendai virus; the product was impure, about one part interferon in a thousand, and in 1979 the laboratory produced 400 billion units a year. The same article says full purification would destroy 99 percent of the interferon and be very expensive.

The third caution is that the clearest test of this story came from a randomized trial. In EURAMOS-1, 2,260 patients were registered; 716 who responded well to preoperative treatment were randomly divided: standard chemotherapy plus pegylated interferon alfa-2b (357) or without it (359). The hazard ratio for event-free survival was 0.83 (P = 0.214); three-year event-free survival was 77 percent with interferon and 74 percent without. In relative terms that looks like a 17 percent lower risk; in absolute terms it is a three-point difference that is not statistically significant, and the authors said the results did not support routine interferon maintenance after standard chemotherapy (Bielack et al., J Clin Oncol 2015, from the ASCO Post summary). This is not the same product or the same patient group as Strander's leukocyte interferon in 1971, so it does not refute his results; it shows only that pegylated interferon as maintenance gave no significant difference in a randomized trial.

The fourth is that the ground interferon held was narrow. In 1984 Quesada and colleagues reported alpha interferon producing complete response in three and partial response in four of seven patients with progressive hairy cell leukemia; the responses lasted more than 6 to more than 10 months (NEJM 1984;310:15-18, from the abstract). It was a very small series with short follow-up. The first approval of Intron A (interferon alfa-2b) came on 4 June 1986, for this disease (hemonc.org). According to Gresser, lung, stomach, colon, prostate and breast cancers were not amenable to interferon alone (1989). In Gresser's own account, mouse experiments began in 1965 with virus-induced leukemias and then extended to transplantable tumors; the mechanism was unclear in either mice or patients.