Extensive disease small cell lung cancer dose-response relationships: implications for resistance mechanisms

Abstract

Background: Some studies (but not others) suggested that high doses are beneficial in small cell lung cancer (SCLC). We hypothesized that dose-response curve (DRC) shape reflects resistance mechanisms.

Methods: We reviewed published SCLC clinical trials and converted response rates into estimated mean tumor cell kill, assuming killing is proportional to reduction in tumor volume. Mean % cell survival was plotted versus planned dose intensity. Nonlinear and linear meta-regression analyses (weighted according to the number of patients in each study) were used to assess DRC characteristics.

Results: Although associations between dose and cell survival were not statistically significant, DRCs sloped downward for five of seven agents across all doses and for all seven when lowest doses were excluded. Maximum mean cell kill across all drugs and doses was approximately 90%, suggesting that there may be a maximum achievable tumor cell kill irrespective of number of agents or drug doses.

Conclusions: Downward DRC slopes suggest that maintaining relatively high doses may possibly maximize palliation, although the associations between dose and slope did not achieve statistical significance, and slopes for most drugs tended to be shallow. DRC flattening at higher doses would preclude cure and would suggest that "saturable passive resistance" (deficiency of factors required for cell killing) limits maximum achievable cell kill. An example of factors that could flatten the DRC at higher doses and lead to saturable passive resistance would be presence of quiescent, noncycling cells.

Figure 1

We postulated that dose-response curve shape would reflect whether resistance was due to deficiency or saturation of a factor required for drug effect (which would result in flattening of the dose-response curve at higher doses, “saturable passive resistance”, analogous to non-competitive inhibition of drug effect), due to mutation, etc resulting in decreased affinity of a drug for an uptake system, target, etc (which would give decreased curve slope, “non-saturable passive resistance”), or due to excess of a resistance factor (which would give a shoulder on the dose-response curve if log effect is plotted against linear dose, “active resistance”, analogous to competitive inhibition of drug effect) . (From Crit Rev Hem Oncol, with permission of the publishers).

Figure 2

Weighted nonlinear meta-regression analysis (where the α parameter measures the relationship between dose and cell survival) or weighted linear regression analysis (where the β parameter measures the relationship between dose and cell survival) over the entire dose range assessed. Data for each of cisplatin (n=30, α = −0.02) and cyclophosphamide (n=8, α = −0.01) could be fitted to the nonlinear regression model, while data for each of etoposide (n=40, β = −0.01), carboplatin (n=11, β = 0.14), epirubicin (n=11, β = 0.25), paclitaxel (n=8, β = −0.11), and ifosfamide (n=11, β = −0.01) could not be fitted to the nonlinear regression model and were instead assessed by linear regression analysis.

Figure 3

Weighted nonlinear meta-regression analysis (where the α parameter measures the relationship between dose and cell survival) or weighted linear regression analysis (where the β parameter measures the relationship between dose and cell survival) over higher doses (with elimination of data for doses of etoposide ≤ 100, cisplatin ≤ 20, epirubicin ≤ 20, cyclophosphamide ≤ 200, paclitaxel ≤ 35 or ifosfamide ≤ 1,250 mg/m2/week or carboplatin ≤ AUC 1.25/week). Data for each of cisplatin (n=22, α = −0.02), carboplatin (n=6, α = −0.39), cyclophosphamide (n=3, α = −0.01), and ifosfamide (n=9, α = −0.0004) could be fitted to the nonlinear regression model, while data for each of etoposide (n=20, β = −0.09), epirubicin (n=7, β = −0.48), and paclitaxel (n=8, β = −0.11) could not be fitted to the nonlinear regression model and were instead assessed by linear regression analysis.