In the study by Martín-Broto and colleagues (1) on biomarkers in patients with soft-tissue sarcomas (STS) treated with doxorubicin plus olaratumab, patients received their first cycle of single-agent olaratumab, which was not in the previously published regimen (2), and only patients with potentially resectable disease were included. Both decisions conflict with the acknowledgment of two confounding effects of small randomized trials with significant crossover: fluctuations in type I error due to random inequalities in the distribution of prognostic variables at baseline (3) and the postrandomization bias associated with the crossover effect (4).

The most frequent bias of the crossover of a very effective experimental agent to the control arm is the obscuring of the overall survival (OS) benefit in the experimental arm (increase in type II error). When the experimental agent is ineffective, the crossover artificially inflates the benefit in OS (increase in type I error). The lower the efficacy of the experimental antitumor agent, the greater the OS benefit for the experimental arm. The small phase II trial of doxorubicin plus olaratumab versus doxorubicin alone (2) showed a huge OS benefit (26.5 vs. 14.7 months in the experimental and control arms, respectively; P = 0.003), with no significant differences in the progression-free survival (PFS; 6.6 vs. 4.1 months; P = 0.06). This result is not “paradoxical” as stated by Martín-Broto and colleagues. It is observed because 46% of the patients in the standard arm crossed over to single-agent olaratumab, which prevented them from receiving more effective second lines. What was “paradoxical” was the approval of the regulatory agencies based on these data.

The median PFS of the 51 patients was 2.86 months with no differences between subgroups (1). However, the authors state that “anti-PDGFR therapy may be effective in a subgroup of patients.” This conclusion is not derived from their results and is different from the results of the ANNOUNCE trial (509 patients), which showed that the PFS of olaratumab plus doxorubicin was significantly shorter than that of doxorubicin alone (5.4 vs. 6.8 months; P = 0.04), with no differences between subgroups (5).

The 2016 failed approval could have been avoided if we had detected that we were fooled by randomness and bias, two powerful enemies of empirical scientific advancement that, like Godzilla and King Kong, can wipe out everything in one fell swoop. The mistakes of 2016 should not be repeated in 2021.

See the Response, p. 2094

P. Jimenez-Fonseca reports honoraria for her consulting/advisory role from Bristol Myers Squibb, Pfizer, Merck Sharp & Dohme, Mylan, Baxter; speaking from HRA Pharma, LeoPharma, Rovi, Sanofi; and travel grants from Ipsen, all outside of the scope of this work. No disclosures were reported by the other authors.

1.
Martín-Broto
J
,
Pousa
AL
,
Brohl
AS
,
Van Tine
BA
,
Powers
B
,
Stacchiotti
S
, et al
Circulating tumor cells and biomarker modulation with olaratumab monotherapy followed by olaratumab plus doxorubicin: phase Ib study in patients with soft-tissue sarcoma
.
Mol Cancer Ther
2021
;
20
:
132
41
.
2.
Tap
WD
,
Jones
RL
,
Van Tine
BA
,
Chmielowski
B
,
Elias
AD
,
Adkins
D
, et al
Olaratumab and doxorubicin versus doxorubicin alone for treatment of soft-tissue sarcoma: an open-label phase 1b and randomised phase 2 trial
.
Lancet
2016
;
388
:
488
97
.
3.
Ziv
G
,
Lidor
R
,
Netz
Y
. 
Dealing with possible baseline inequalities between experimental groups - the case of motor learning
.
J Mot Behav
2020
;
52
:
502
13
.
4.
Fojo
T
,
Amiri-Kordestani
L
,
Bates
SE
. 
Potential pitfalls of crossover and thoughts on iniparib in triple-negative breast cancer
.
J Natl Cancer Inst
2011
;
103
:
1738
40
.
5.
Tap
WD
,
Wagner
AJ
,
Schöffski
P
,
Martin-Broto
J
,
Krarup-Hansen
A
,
Ganjoo
KN
, et al
;
ANNOUNCE Investigators
. 
Effect of doxorubicin plus olaratumab vs. doxorubicin plus placebo on survival in patients with advanced soft tissue sarcomas: The ANNOUNCE Randomized Clinical Trial
.
JAMA
2020
;
323
:
1266
76
.