Cooperative group protocols giving 3 days of anthracycline and 7 of cytarabine (3+7) to patients with newly-diagnosed AML typically call for a 2nd 3+7 in patients not in CR after course 1. However, it is unclear how often this practice is followed and, in particular, whether patients who receive a 2nd course differ from those who do not. If so, results following the second course might be confounded by selection bias, as would results of any “salvage” therapy given to patients taken off after course 1.


Our database consisted of 595 patients aged 60 with de novo AML given 3+7+/- gemtuzumab ozogamicin (GO) on SWOG protocol S0106 (Blood 2013 121(24):4854-4860). 237 (40%) were not in CR after course 1, of whom 19 died on this course, leaving 218 who per protocol should have received a 2nd 3+7. Information is available on 216 of the 218. We compared patients who did and did not receive a 2nd course with respect to age, marrow blasts on “day 14” (the median when the first post treatment exam was performed), day 14 cellularity, day 14 infiltrate (blasts X cellularity), and change in blasts, cellularity and infiltrate vs. pre-treatment. Each of these covariates were regarded numerically and we also examined the categorical covariates pre-treatment performance status (PS, 0-1 vs. 2-3), cytogenetics (SWOG criteria), and whether treatment was at a SWOG member institution or a SWOG non-member institution (with intergroup patients registered through the CTSU excluded from associated analyses). We used multivariate analysis to identify covariates independently associated with receiving a 2nd 3+7 and area under receiver operating characteristic curves (AUC) to quantify how effective the multivariate models were at predicting who received a 2nd course (AUC 1.0 = perfect prediction, AUC 0.5 = “coin flip”).


119 of the 216 patients (55%) not in CR after course 1 received course 2, and this proportion was the same for patients given or not given GO on course 1. 76% of the 97 patients who went off protocol received alternative chemotherapy, 6% received an allogeneic transplant, 7% received no further therapy, and in 10% information was lacking. Patients who received a second course had a higher day 14 blast % than off-protocol patients (p = 0.005, medians 28% vs. 15%), but did not differ in day 14 cellularity, infiltrate, or change in any of these parameters vs. pre-treatment. Off protocol and 2 course patients were similarly aged (medians 47 and 48). A 2nd 3+7 was given to 59% of patients with unfavorable cytogenetics and 57% of others. 58% of 186 patients with PS 0-1 received a 2nd course vs. 41% of 29 with PS 2-3 (p=0.11). 50% of the patients treated at SWOG member sites went off protocol vs. 32% at non-member sites (p=0.05). However, characteristics of patients who went off protocol were similar at member and non-member sites, with the exception that patients with unfavorable cytogenetics comprised a higher proportion of patients who got further therapy (on or off protocol) at member sites (p=0.03). Multivariate analysis indicated that being a SWOG member site was the only covariate that was independently associated with a patient receiving a 2nd course , with non-member patients being 6.1-fold more likely to do so (p=0.014). The AUC for the multivariate model was 0.70.


The AUC from the multivariate model suggests that our ability to identify patients who will receive a 2nd course of 3+7 rather than being removed from protocol is quite limited. Within this constraint, patients treated at SWOG member sites are more likely to be removed from protocol treatment. Whether this simply reflects the availability of more therapeutic options or a more systematic bias remains to be determined. Identifier: NCT00085709


This work was supported in part by PHS Cooperative Agreement grants awarded by the National Cancer Institute, CA32102 and CA38926, and in part by Wyeth (Pfizer) Pharmaceuticals.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution