Relationship between diagnosis-specific activity of cytotoxic drugs in fresh human tumour cells ex vivo and in the clinic

Abstract

The aim of this investigation was to evaluate the relationship between the disease-specific activity of cytotoxic drugs in the clinic and in fresh human tumour cells from patients, as detected by a non-clonogenic cytotoxicity assay. The activity of 18 different cytotoxic drugs in fresh human tumour cells ex vivo was analysed in up to 15 samples from each of 13 different diagnoses using the fluorometric microculture cytotoxicity assay (FMCA). For each drug and diagnosis, relative activity indices (RAIs) were calculated, defined as the fraction of samples within the diagnosis having a survival index (SI) less than the median SI for the drug in all tested samples. Clinical response rates for the drug in the same diagnoses were collected from published phase II trials and were compared with the RAIs using Spearman's rank correlation (Rho) and Pearson's correlation coefficient (R). Correlation coefficients could be calculated for 12 drugs. The coefficients varied between 0.02 and 0.92 (Rho) and between 0.07 and 0.91 (R), but for most drugs the correlation between RAI and clinical response rates was good, with Rho >0.6 and R>0.7. Weak correlations were observed for cyclophosphamide and vincristine (Rho=0.32 and 0.16, respectively), which might be due to old clinical data, and for paclitaxel (Rho=0.02), which perhaps could be explained by in vitro activity of the solvent, Cremophor EL. The 18 drugs were also categorised according to their suggested clinical use in solid or haematological tumours, and were then compared regarding the activity in solid compared with haematological tumours ex vivo, expressed as the ratio between the number of responders among solid and haematological tumours (S/H ratio). The mean ex vivo S/H ratios in the group of drugs registered for use in haematological tumours was only 0.09 and was significantly different (P=0.05) from the mean S/H ratios for the drugs used in both haematological and solid tumours (0.31) and in solid tumours only (0.47). Furthermore, the FMCA could identify the 50% most and least sensitive diagnoses with respect to clinical phase II activity with 74% (78/106) correct classifications. The results indicate that the relative activity of cytotoxic drugs in different diagnoses may be detected by the FMCA. Thus, ‘phase II trials ex vivo’ using non-clonogenic cytotoxicity assays might be used to make clinical trials more effective by targeting trials to diagnoses in which a new agent is most likely to be active.

Introduction

The means to determine the activity of new cytotoxic drugs in different tumour types has thus far been the empirical phase II clinical trial. This approach involves financial as well as ethical problems, since there is a risk of giving a new agent to a large number of patients with diagnoses in which the drug turns out to be ineffective[1]. A valid method for preclinical evaluation of disease-specific activity of new anticancer agents could, therefore, be of value in the planning of clinical trials, by directing the phase II trials into the specific tumour types in which the new agent is most likely to be effective.

During the few last decades, a number of laboratory assays for the determination of drug resistance in fresh tumour cells from patients ex vivo have evolved[2]. The primary objective of the development of these assays was to tailor the individual patient’s treatment. We have developed a non-clonogenic fluorometric microculture cytotoxicity assay (FMCA) which is a rapid and efficient method, applicable to most tumour types3, 4. The feasibility of the FMCA for prediction of individual cytotoxic drug sensitivity is well documented. The assay has so far been applied to more than 2500 patient tumour samples, and the specificity and sensitivity of the test for prediction of individual tumour response are in the same range as for other non-clonogenic cytotoxicity assays, such as the MTT and differential staining cytotoxicity (DiSC) assays5, 6, 7. Much less is known about the ability of the FMCA and other assays to detect disease-specific activity. Although some studies have indicated this potential8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, no quantitative estimates of this relationship have been reported.

The aim of the present study was to evaluate the feasibility of the FMCA for prediction of cytotoxic drug activity in different diagnoses, by performing statistical analyses of the relationship between the activity pattern of cytotoxic drugs as determined in primary cultures of human tumour cells using the FMCA and as observed in the clinic. To obtain this, we used a defined number of patient samples for the determination of an activity index ex vivo and phase II clinical response rates as a numerical measure of clinical activity.