Background: The rest-activity rhythm is a biomarker of the circadian timing system, whose disruption both accelerates cancer progression in experimental models (Filipski et al. JNCI 2002, 2005) and predicts for poor overall survival in cancer patients (pts) (Innominato et al. Cancer Res 2009; Int J Cancer 2012). Here we investigate the relevance of early signals of circadian disruption for the antitumor efficacy of intravenous Cet and chronomodulated triplet HAI in pts with unresectable and refractory LM-CRC. Methods: Rest-activity was monitored q1min using a wrist watch actigraph worn for three weeks, starting 1 week before HAI onset. A robust and validated parameter (I<O), the relative measure of activity In-bed vs Out-of-bed, was used to evaluate rest-activity patterns. Tumor response was evaluated using RECIST criteria. Results: 10M and 3F, aged 33-72 y (median, 60 y) had a good Performance Status (0, 10 pts; 1, 3 pts) and received one (7 pts), two or three (6 pts) prior chemotherapy lines. Pts displayed complete or partial response (1 CR, 6 PR), or stable disease (SD, 6 pts). Median I<0 was 95.4% (25-75% quartiles, 88.8-98.0) at baseline, 92.8% (89.4- 96.2) during the treatment week (p<0.001 vs “baseline”) and 95.0% (89.2- 97.6, NS) on the week after HAI. On this last week, I<O improved or recovered fully in 5 pts (38.5%), or partly in 3 pts (23.1%). Rest-activity rhythm disruption persisted in 5 pts (38.5%). An objective tumor response was achieved in 7/ 8 pts whose I<O improved or recovered (87.5%) while the poor I<O computed for 5 patients on the week after HAI predicted for SD only (p=0.003). This relation was independent of PS and prior chemotherapy lines (p=0.013). Conclusions: The robustness of the rest-activity rhythm during the first course of Chrono-Optiliv was an early signal that predicted subsequent antitumor activity. Further testing of this circadian biomarker is warranted as an early surrogate endpoint of antitumor efficacy of Chrono-Optiliv. Clinical trial information: NCT00852228.