In 1984 Mackillop et al.1 sent a questionnaire to physicians and surgeons in Ontario who treated lung cancer. The questionnaire asked the doctors to consider themselves as patients with lung cancer, as specified in a number of clinical scenarios, and to describe the treatment they would wish to receive. One scenario described asymptomatic, locally advanced non-small-cell lung cancer. A majority of the respondents (61 percent) preferred treatment with radiotherapy, and 22 percent preferred no immediate treatment; only 5 percent of the respondents chose chemotherapy. We do not know the reasons for these choices, but they were presumably influenced by randomized trials (reviewed by Payne2) that showed at most small benefits in survival when immediate radiotherapy was compared with a wait-and-see policy, and by six randomized trials of radiation with or without chemotherapy (reviewed by Tannock3), only one of which showed a small survival advantage for combined treatment as compared with radiation alone. The doctors were expressing the opinion that if they had cancer of a type and stage almost certain to be lethal, the side effects associated with drug treatment were not worth enduring. Has the situation changed in 1990?
In this issue of the Journal, Dillman et al.4 present the results of a clinical trial in patients with locally advanced non-small-cell lung cancer who had excellent performance status. There was a small but significant difference in the probability of survival favoring the patients randomly assigned to receive a five-week course of induction chemotherapy (cisplatin and vinblastine) before radiotherapy, as compared with radiotherapy alone. There was an approximate doubling in the number of long-term survivors (projected to be 23 and 11 percent, respectively, after three years). Nonetheless, most patients had died of their disease, and only 25 of 155 eligible patients (16 percent) were alive without progressive disease when the report was written. The trial was conducted carefully, and the investigators are to be commended both for choosing a group of patients who were most likely to benefit from added chemotherapy and for their inclusion of an economic analysis. Should combined therapy now be offered to such patients as standard treatment? An affirmative answer to this question depends both on confidence that the results of the trial can be generalized and on the judgment that the limited benefits of combined treatment are worthwhile when balanced against the toxicity and cost associated with its administration.
With regard to the general validity of the trial, there is obviously no basis for extending the results to patients with similar stages of disease whose performance status is poorer: Dillman et al.4 estimated that about 25 percent of all patients with Stage III lung cancer would fit the eligibility criteria of their trial. Furthermore, there is some concern about the degree of selection of patients who took part in the trial. The patients were enrolled from 22 institutions over a three-year period, which represents a mean rate of accrual of fewer than three patients per year per institution. This is a low rate of accrual for a common disease. A record of eligible patients seen but not enrolled would be useful, since a trial may not produce representative results if only a small proportion of eligible patients is enrolled. Finally, as the authors recognize, a confirmatory trial is important. Seemingly significant results can occur by chance alone, and the results of two similar trials now being conducted by other cooperative oncology groups should allow an assessment of the general validity of the results reported by Dillman et al.
In a variety of types and stages of cancer, physicians and their patients commonly face choices between treatments that lead to small differences in outcome. Examination of the survival curves presented by Dillman et al. suggests that for every patient who benefited from combined treatment, about seven others were exposed to the cost and toxicity of treatment with no evidence of gain. In an age of diminishing health care resources, is it justifiable to recommend a therapy with such poor odds of success? The answer to this question will depend on the perspective from which it is viewed: judgments about the balance between therapeutic gain and toxicity or cost will vary among patients, physicians, and those who pay for health care. For third-party payers with limited budgets, the issue becomes one of comparing the cost and efficacy of a treatment with what could be achieved by using the same resources to treat another disease.
There is no absolute measure of benefit that can be used to determine whether a particular treatment is worthwhile. The traditional end points in trials of cancer treatment are tumor response and duration of survival. Although the shrinkage of tumor (i.e., response) may give information about the biologic activity of the therapy, it does not necessarily mean benefit to the patient. Balancing small gains in survival against toxicity and cost has been particularly difficult, because they are fundamentally different considerations, like the proverbial apples and oranges. The more recent inclusion of quality of life as an end point in clinical trials of cancer therapy has provided a way to incorporate both positive outcomes (the relief of symptoms) and negative outcomes (the toxicity of treatment) into a measure of the overall value of a treatment. One thereby attempts to assess the quality of the apples and oranges, as well as count them. Indexes of the quality of life have now been developed and validated for use in patients with several types of cancer. They can be applied when the primary aim of treatment is to provide palliation5 or, in the setting of adjuvant chemotherapy, when the aim is to improve survival or delay relapse.6
Complementary methods are being developed to estimate the differences in therapeutic outcome that patients (or their surrogates) regard as worthwhile. With trade-off methods, a patient is asked to choose between two potential outcomes. An example is the choice between a defined period of poor health due to treatment, followed by prolonged survival, and shorter survival with no immediate ill effects of treatment. The difference in survival (other end points can also be used) can then be varied until the patient reaches a point of indecision.7 Among patients with operable lung cancer, the method revealed that an appreciable number preferred radiation therapy to surgery (thus avoiding the risk of perioperative mortality), despite a lower probability of long-term survival. In contrast, among patients who have received chemotherapy for breast cancer, the method suggested that they would accept toxic treatment for rather small gains in survival.8 , 9
The economic cost of treatment is becoming increasingly important in selecting therapy, and the rationing of marginally effective treatments has been proposed as a way to contain the costs of medical care.10 Small gains in the duration or quality of survival probably cannot be justified if they are very expensive. Economic analyses, however, sometimes produce surprising results. Although it requires confirmation, an analysis of a trial comparing chemotherapy with the best supportive care for metastatic non-small-cell lung cancer showed that the additional cost of the drugs could be outweighed by the savings from less frequent hospitalization.11 Chemotherapy does not always add expense, but it usually adds the burden of toxic side effects.
The economic and toxicity analyses provided by Dillman et al.4 support the conclusion that combined treatment was cost effective in their patients, both in dollars and in side effects. Provided that the results of this trial are reproducible, it would seem appropriate to discuss the option of combined therapy with otherwise healthy patients who have locally advanced, inoperable lung cancer.
