Dose-response relationship for prophylactic cranial irradiation in small cell lung cancer

Purpose: To determine the dose-response relationship for prophylactic cranial irradiation (PCI) in small cell lung cancer, to quantify the growth kinetics of subclinical metastases, and to determine the influence of time-delay in initiating PCI on its utility, Methods and Materials: Published reports of brain relapse rates in small cell lung cancer with and without PCI were collected, The reduction in brain relapse rate as a function of radiation dose was analyzed, The time interval between treatment of the primary tumor and the initiation of PCI was analyzed as a factor potentially influencing dose-response, Results: A shallow dose-response curve without any threshold in the dose intercept was demonstrated for control of subclinical brain metastases in "early PCI" (delay between initiation of treatment for primary tumor and PCI less than 60 days), By contrast "late PCI" (delay over 60 days) was associated with a significant displacement of the dose intercept, Doses over 30-35 Gy in 2-Gy fractions did not result in a further reduction in brain relapse rate, but there were too few high-dose studies to draw any definite conclusion, Conclusions: The nearly linear dose-response relationship for reduction in brain relapses demonstrated for "early PCI" in the range of doses from zero up to 35 Gy given in 2-Gy fractions supports the model of a fairly logarithmically uniform distribution of metastatic cell number within a series of patients, When PCI is delayed, a significant threshold in dose-response was 'observed, consistent with a fast growth rate of untreated subclinical brain metastases from small cell lung cancer, The exact shape and locations of dose-response curves is not well established by this retrospective analysis of diverse data, A high probability of eliminating brain relapses following PCI requires a dose of about 30-35 Gy in 2-Gy fractions, Control rates in brain can be enhanced if PCI is applied early. (C) 1998 Elsevier Science Inc.