5‐year cure rate: Yet another myth
S. Friberg
DOI: https://doi.org/10.1002/(SICI)1096-9098(199706)65:2<73::AID-JSO1>3.0.CO;2-K
1997-06-01
Journal of Surgical Oncology
Abstract:The 5-year survival rate has long been cited as an index of the effectiveness of cancer treatment. It has been in wide use since the end of the nineteenth century, included in thousands of scientific publications. Basically, the 5-year survival rate is a statistical tool for characterizing the survival of a certain group of individuals. It is somewhat primitive [1,2], since it does not utilize all available information from a survival analysis. It tells us only how large a proportion has survived for a limited time-span—arbitrarily put to 5 years. It is not even a true rate, but simply a value at a fixed point of time [3]. In fact, it is easy to give examples of quite different survival patterns with identical 5-year survival rates (Fig. 1). If the original intention with the ‘‘5-year survival’’ had been maintained—as a mere value in time—no harm would have been done. But over the years a change has occurred in the meaning—interpretation?—of the term ‘‘5-year survival rate.’’ It has gradually shifted to become the equivalent to ‘‘5-year-cure rate.’’ This is a gross misconception, because ‘‘survival’’ is not synonymous with ‘‘cure.’’ Only for fast-growing tumors (e.g., acute leukemias and testicular nonseminomatous germ cell tumors) is surviving for 5 years after the diagnosis likely to indicate a true cure. This is illustrated by curve B in Figure 1. Most human malignancies, however, are slow-growing [4,5], i.e., most cancers of the breast, colon, or prostate, and require many years or even decades to kill their host. For such cases, survival for 5 years is not indicative of cure. For example, of the women who survive their breast cancer diagnosis for 5 years, 1/3 will succumb to their disease, and for those women who survive for 10 years, as much as 1/4 will still die from their malignancy. Even as long as 30 years after the diagnosis, this patient population shows an excess mortality from cancer of the breast. Consequently, some clinicians have posed the question: Do we ever cure cancer of the breast? [6–21]. Apart from being arbitrary and providing only curtailed information, any limited observation period (such as 5 years) also has the disadvantage of allowing one dditional pitfall: ‘‘lead time bias’’ [22–24]. If the time of diagnosis is shifted to an earlier date, the fixed observation period will automatically lead to fewer observed deaths at the end of that period. The impression is, therefore, given that ‘‘early diagnosis leads to better prognosis.’’ This conclusion is erroneous: Shifting the time of diagnosis of a patient to an earlier date means only that the patient (if untreated) will live longer as a cancer patient, although not as an individual. The cliché‘‘earlier diagnosis leads to better prognosis’’ and the misconception ‘‘5-year cure’’ have done more harm than good. The cliche ́ lacks firm scientific support, and the words ‘‘early’’ and ‘‘cure’’ are often used without being defined. If ‘‘early’’ is supposed to mean ‘‘before the tumor has metastasized,’’ this is another illusion. Most human tumors have metastasized when diagnosable by present-day methods [25–30]. Further, the word ‘‘cure’’ should be used with caution. Many adult malignant tumors have a long natural course extending over two or three decades [4,5]. This means that there are many individuals who, after having had their primary tumors removed, live with asymptomatic, microscopic disease (like many other chronic diseases). These patients have not been cured, but they live seemingly healthy lives with ‘‘no evidence of disease.’’ In clinical oncology, a wholly satisfactory definition of the term ‘‘cure’’ is, therefore, regarded as difficult—if not almost impossible [19]. There are three current definitions of cure: (1) statistical, (2) clinical, and (3) personal. ‘‘Statistical cure’’ means that the study population dies at the same rate as the ‘‘normal’’ population (regardless of diagnosis). ‘‘Clinical cure’’ designates the