904 / 1851
The study did not meet many standards…
in design, undertaking, and interpretation.
Poor protocol adherence
(318/551 = 57.7%)
Low statistical power
because of exclusion of patients with protocol violations from the
denominator in various analyses (PP vs ITT analyses)
Long accrual period
(2000-2009, 75 centers)
Involvement of many centers with little experience in PCNSL
(high rates of
erroneous response assessment, patients lost to follow-up, and toxic deaths).
Inconsistent data for iatrogenic neurotoxicity
(exclusively assessed by MRI)
The G-PCNSL-SG-1 trial is compromised by several
flaws and failed to prove its primary hypothesis.