Analysis of prognostic variables in hyperthermia treatment of 161 patients.

From 1977-1982, 161 patients were treated using hyperthermia as an adjuvant in Phase I trials. Microwave applicators (MW), capacitively coupled plates (RF plates), interstitial localized current fields (LCF), and magnetic induction heating (MI) techniques were used together with radiation in 135 patients, with chemotherapy in 10 patients, and alone in 16 patients. Tumor volume response categories were no response (NR, less than 50% decrease); partial response (PR, 50% less than or equal to volume decrease less than 100%); and complete response (CR, complete disappearance). The CR rates and total response rates (CR + PR) were 38/160 (24%) and 90/160 (56%), respectively. There were highly significant differences among techniques in CR vs PR + NR (p = .001), and in CR + PR vs NR (p less than .0005). Response did not vary significantly with histologic category. Overall toxicity was 16%, and did not vary significantly with technique (p = .193). In the patient group treated with hyperthermia and radiation, multivariate analysis revealed that a set of three variables had prognostic importance for CR: technique (p = .011), radiation dose (p = .019), and tumor volume (p = .001, negatively correlated). A good correlation also existed between CR and the minimum tumor temperature averaged over all treatments, TMIN (p less than .0005). Temperature variables themselves were correlated with tumor volume. Minimum T correlated negatively with volume (p = .017) and TMAX correlated positively with volume (p = .026). In fewer than 50% of patients could minimum T greater than 40.7 degrees C be achieved. Our conclusions are: TMIN, tumor volume, radiation dose, and heating technique have prognostic value for initial response; variation in CR vs technique reflects variation in tumor volume treated and in minimum temperature achieved with these techniques; and acute toxicity of treatment is infrequent, but serious toxicity is possible with the interstitial technique.