Cataract development: it's cumulative.

The effects of cell density on drug-induced cell kill kinetics were studied by means of clonogenic assay using 3 human leukaemia-lymphoma cell lines. Mitoxantrone, daunorubicin, doxorubicin, vincristine and bleomycin were progressively less efficacious when cell density increased (positive inoculum effects), whereas the effects of cis-platin and carboplatin were not influenced by cell density. Inoculum effects were related to the kind of chemotherapeutic agents tested, irrespective of the type of cell lines used. Preincubation of mitoxantrone or doxorubicin in the presence of cells in high density resulted in decreases in the cytocidal activity, whereas the effects of bleomycin, vincristine and cis-platin were unaffected. These results show that cell density affects the biological effect of certain chemotherapeutic agents. Inactivation of drugs by high densities of cells partially explains this phenomenon. For many years the relationship between tumour size and sensitivity to anticancer agents affected chemotherapeutic strategies (Skipper & Schabel, 1982). Clinical experience suggests that patients with high tumour load are less amenable to treatment. High leukocyte counts in patients with ALL and AML are usually considered a poor prognostic sign in terms of remission induction Leukapheresis prior to chemotherapy resulted in an improved rate of remission (Cuttner et al., 1983). These reports suggest that not only the total tumour load, but cell density per se is an important factor for chemotherapeutic efficacy. We studied the effects of cell density on drug-induced cell kill kinetics in vitro.

For many years the relationship between tumour size and sensitivity to anticancer agents affected chemotherapeutic strategies (Skipper & Schabel, 1982). Clinical experience suggests that patients with high tumour load are less amenable to treatment. High leukocyte counts in patients with ALL and AML are usually considered a poor prognostic sign in terms of remission induction (Crowther et al., 1975;Robinson et al., 1980;Simone et al., 1975), remission duration (Hug et al., 1983;Robinson et al., 1980;Schauer et al., 1983) and survival (Freireich et al., 1961;George et al., 1973;Simone et al., 1975). Leukapheresis prior to chemotherapy resulted in an improved rate of remission (Cuttner et al., 1983). These reports suggest that not only the total tumour load, but cell density per se is an important factor for chemotherapeutic efficacy. We studied the effects of cell density on drug-induced cell kill kinetics in vitro.

Cell lines
The DND-39A Burkitt lymphoma cell line was derived from the pleural effusion of a man with American Burkitt lymphoma (Ohnuma et al., 1980). The MOLT-3 T-lymphocyte leukaemia line was established from a patient with ALL (Minowada et al., 1972). The HL-60 acute promyelocytic leukaemia cell line (Gallagher et al., 1979) was a gift from Dr. Ruscetti of the National Cancer Institute, Bethesda, MD. These cell lines were maintained in our laboratory as a suspension in culture flasks containing RPMI-1640 medium (GIBCO, Grand Island, NY) containing 10% (v/v) heat-inactivated FBS (GIBCO, Grand Island, NY) and fed 3 times per week with fresh medium. Drug powders were reconstituted initially according to the accompanying instructions and further dilutions in PBS. The drug dilutions were 'freshly prepared for each experiment.
Determination of drug-induced cell kill kinetics Exponentially growing cells (2-10 x 105 viable cells ml -1) were diluted or concentrated to different cell densities ranging from 102 to 108 viable cellsml-1 in fresh RPMI-1640 medium containing 10% FBS in either 15ml culture tubes (Falcon No. 3033) or 50ml centrifuge tubes (Falcon No. 2098). The cells were exposed to graded concentrations of each drug for 1 h at 37°C in humidified 5% C02/95% air. Drug exposure was carried out in a total volume ranging from 1 ml (108 cells) to 200 ml (102 cellsml-1). At the end of the incubation period the cells were washed twice with medium Correspondence: T. Ohnuma Received 20 February 1986; and in revised form, 9 May 1986. 416 T. OHNUMA et al. and cell densities were adjusted to 3 x 104 cellsml-1 for DND-39A and 5 x 104 cells ml-1 for MOLT-3 and HL-60 cells. Aliquots (100pl) of cell suspension were plated on 0.5% agar (Noble agar, Difco, Detroit, MI) replica plate (60 x 15 mm, Corning, NY) containing RPMI-1640 medium and 10% FBS (Koroki, 1970). Colonies (more than 40 cell aggregates) were enumerated and the survival fraction was determined. In this system the number of colonies developed is proportional to the number of cells plated. When colonies number more than 2,000-3,000/plate, however, they conglomerate, thus, making it difficult to count them accurately. The plating efficiency of DND-39A cells was 30-40%, and that of MOLT-3 and HL-60 cells, 10-20%. The plating efficiency was unchanged after 1 h of incubation in control cells with no drug within a range of cell densities studied. Each experiment was done in triplicate and repeated at least twice.
Inactivation studies For inactivation studies each drug at a concentration equivalent to 10 times the LD30_50 for 104 viable DND-39A cells mlwas preincubated in the medium alone, in the medium containing 104 cells ml -1 or 108 cells ml -1 for l h at 37°C in a humidified 5% C02/95% air incubator. The cells were centrifuged and the supernatant was separated. The 300,1 aliquots of supernatant were added to 3 ml of cell suspension containing 104 DND-39A cells ml1 for 1 h. The cells were then washed free of drug and the cell kill effects were determined by clonogenic assay. Drugs with no preincubation were also included as controls.

Results
The influence of cell density on the drug concentration-cell lethality curves (dose-response curves) for 7 diverse chemotherapeutic agents is shown in Figure 1. As the DND-39A cell density increased, the efficacy of mitoxantrone, doxorubicin, daunorubicin, bleomycin and vincristine progressively decreased and the dose-response curve moved to the right. This decrease in efficacy was most pronounced with mitoxantrone ( Figure IF).
Thus, the LD50 of mitoxantrone was 1.4, 1.8, 6.0 and 126-fold higher when the cell density increased from 102 to 104, 106, 107 and 108 cellsmlP-. When cell density increased from 104 and 108 cells mlthe changes in LD50 were in decreasing order, 30-fold for daunorubicin, 27-fold for doxorubicin, 15-fold for vincristine and 9-fold for bleomycin. In contrast, the cell kill effects of cisplatin and carboplatin were independent of cell density and the dose-response curves for each cell density studied overlapped one another (Figure lB and IC). The influence of cell density on the cell kill effects of certain chemotherapeutic agents is similar to the effect of the size of microbial inoculum in vitro for certain antimicrobial agents (see Discussion) and thus can be termed 'inoculum effect'.
In order to examine whether the inoculum effect described above in certain chemotherapeutic agents seen with the DND-39A cell line was specific for this particular cell line we tested the effect of cell density using 2 other cell lines, MOLT-3 and HL-60, and the results were compared between mitoxantrone and cis-platin ( Figure 2). Mitoxantrone, again, showed positive inoculum effect for these 2 cell lines and the dose-response curves for all 3 cell lines moved to the right when the cell density increased from 104 to 108 cells ml-1. In contrast, for all the 3 cell lines tested, cis-platin showed a lack of inoculum effect and there was no shift of the dose-response curves whether the cell density was 104 cells mlor 108 cells ml-1.
One plausible explanation for the positive inoculum effect is the progressive inactivation of the cytocidal effect of these chemotherapeutic agents by the tumour cells. In order to test this hypothesis, the effect was evaluated after preincubation of the drugs with 108 DND-39A cellsml-l and the results are shown in Figure 3. Preincubation of mitoxantrone and doxorubicin with the high density of cells resulted in substantial decreases in activity, whereas the efficacy of bleomycin, vincristine and cis-platin were unaffected.

Discussion
The present study clearly establishes cell density as an important factor in the biological activity of certain chemotherapeutic agents. In the realm of antimicrobial antibiotics this phenomenon is known as the inoculum effect; increases in the inoculum size of certain microbial organisms produce decreases in the area of antimicrobial effects (Basker et al., 1979;Bodey & Pan, 1977).
The degree of inoculum effect was variable for different chemotherapeutic agents tested. Thus, by comparing the ratio of LD50 at 108 cells ml-1 over that at 104 cellsml-l the inoculum effect was most pronounced with mitoxantrone, followed, in the decreasing order, by daunorubicin, doxorubicin, vincristine, bleomycin, cis-platin and carboplatin. In using 3 different cell lines we have demonstrated that the inoculum effect is not related to the type of cells tested, but to the kind of drugs used.  Figure 1 The influence of cell density on drug concentration-cell lethality. curve. DND-39A Burkitt lymphoma cells at different cell densities were exposed to various chemotherapeutic agents for 1 h, washed free of drug and cell lethality was determined by clonogenic assay. The numbers next to each curve indicate cell density (number of cellsml-l of culture medium). Each data point represents a mean of at least 6 experimental values. All experimental values were within 50% of the mean. We studied antibiotics (bleomycin, daunorubicin and doxorubicin); synthetic aminoanthraquinones (mitoxantrone); vinca alkaloids (vincristine) and platinum containing compounds (cis-platin and carboplatin). The mechanisms of action included scission of DNA (bleomycin), intercalation with DNA (daunorubicin, doxorubicin, mitoxantrone and platinum analogues) and tubulin toxins (vincristine) (Calabresi & Parks, 1985). The observation of inoculum effects in diverse chemotherapeutic agents in terms of chemical structure and mechanism of action indicates that this phenomenon is common among clinically active chemotherapeutic agents.
We did not test anti-metabolites because a preliminary study showed that 5-fluorouracil was not active against DND-39A cells in 1 h exposure experiments, up to 1 x 103 M concentrations. Cytosine arabinoside also required more than 1 h of drug exposure to be active. Similarly, we were unable to evaluate melphalan, an alkylating agent, mainly because of its instability in solution.
The mechanism of the inoculum effect produced by certain anticancer agents is unclear. The preincubation studies have only partially explained this phenomenon. Preincubation studies should be positive if drugs are either inactivated by or absorbed into the cells. If drugs are simply adsorbed into the cells they should exert the same degree of lethal effect, irrespective of cell density. Since this was not the case for mitoxantrone and doxorubicin (Figure 1) it is more likely that the drugs were inactivated during the preincubation period. (1 x 10-5M) Figure 3 Effects of preincubation of chemotherapeutic agents on DND-39A cell lethality. Each drug at a concentration equivalent to 10 times the LD30 50 for 104 cells ml-cell density was preincubated for h in the medium alone (bar B), in the culture medium containing 104 cells ml-(bar C) and in the culture medium containing 108 cells ml -(bar D). The supernatant was separated and its cell kill effect was evaluated by adding in a new tube containing DND-39A cells. Bar A indicates the efficacy of each drug freshly prepared and evaluated in the same manner. Error bars, s.d.
This interpretation is reminescent of that seen with certain antimicrobial agents. For example, ureidopenicillins' inhibitory concentrations in vitro were shown to be influenced by the size of the inoculum of Klebsiella aerogenes and Pseudomonas aeruginosa, and this phenomenon was reported to correlate with the degree of inactivation of ureidopenicillins by ,B-lactamases from these microorganisms (Basker et al., 1979).
Inactivation (or adsorption) of chemotherapeutic agents by tumour cells could not explain the total phenomenon, however. Thus, bleomycin and vincristine, both of which showed a positive inoculum effect, were not influenced by preincubation with the cells. In these cases the following alternative explanations can be offered: First, concentrations needed for the doseresponse curve for those with positive inoculum effects, were, in general, lower than cis-platin and carboplatin on a molar basis. It is conceivable, therefore, that there was not enough drug available to the receptor or binding sites of all the cells when high cell densities were used, which might have resulted in more cells remaining unaffected. Since platinum analogues and anthracycline antibiotics are known to intercalate with DNA, DNA is probably the major binding site for these compounds. They also bind with cell membrane components and certain anthracycline compounds with cytoplasmic components (Young et al., 1981). Tubulin proteins are known to be specific receptors for vinca alkaloids (Totsuka et al., 1982;Wilson et al., 1974).
Second, it is possible that in the cases of bleomycin and vincristine, cells are not uniformly sensitive to these agents and that a large inoculum size might have included more cells that are inherently resistant to these agents. Less steep doseresponse curves for bleomycin and vincristine suggest that this alternative mechanism exists. Such an explanation has been advanced for the inoculum effect observed with mezlocillin (Bodey & Pan, 1977).
The present study shows that the relationship between the size of microbial inoculum and the efficacy of antimicrobial antibiotics seems to exist for the size of tumour cell concentrations and the biological effects of certain anticancer agents. It is uncertain at this time whether the lack of inoculum effect seen with cis-platin and carboplatin is specific for platinum analogues or for other agents with similar mechanisms of action, including alkylating agents.
Our observation may have a bearing in several important areas of cellular pharmacology. First, human tumour clonogenic assays are reported to be useful in predicting response (Salmon et al., 1978;von Hoff et al., 1981). In such an assay usually 105 cells ml-I are exposed to anticancer agents. Observations made in this study suggests that 105 cells used in in vitro assay may not represent the circumstances in vivo for certain drugs and may over-predict for positive effects. Second, it seems probable that drugs with high inoculum effects (e.g., mitoxantrone and daunorubicin) are thosewith lesser activity against solid tumours. Determination of inoculum effects may be useful in deciding which drugs are therapeutically active for patients with solid tumours. Third, the present study indicating that certain drugs are less efficacious at high tumour cell densities may partly be a reflection of known in vivo drug effects, in that chemotherapeutic agents are often more active for micrometastatic diseases, than bulky disease. Finally, the lack of inoculum effect of cis-platin and carboplatin seems consistent with their usefulness in the treatment of patients with solid tumours. This observation has now been extended to studies involving multicellular tumour spheroids (Inoue et al., 1985).