Background: NGR-hTNF is a VTA consisting of TNF-α fused to the tumour-homing peptide NGR, which is able to selectively bind an aminopeptidase N overexpressed on tumour blood vessels. Methods: CRC pts failing standard therapies received low-dose NGR-hTNF given at 0.8 μg/m2 as 1-hour infusion every 3 weeks (q3w; triweekly cohort). Progression-free survival (PFS) was the primary study objective with restaging performed q6w. A 2-stage design was used, with 16 and 27 pts to be recruited. Ultimately, an additional 13 pts were treated with 0.8 μg/m2 on a weekly basis (weekly cohort). Results: In the triweekly cohort, 111 cycles (range, 1-10) were delivered to 33 pts with radiologically-documented progression after last therapy. Pts characteristics were: median age: 65 years (range, 5379); M/F 16/17; PS 0/1 26/7. Median number of prior lines was 3 (range, 2-5), whereas 8 pts (25%) had received ≥4 lines and 22 (67%) biologicals. No grade 3-4 drug-related toxicity was observed. Predominant grade 1-2 toxicities were short-lived, infusion-related chills (53%). The median PFS was 2.5 months (95% CI, 2.2-2.8) and the PFS rates at 3 and 4.5 months were 31% and 16%, respectively. The disease control rate was 39% (95% CI, 23-55), with one partial response (3%) and 12 stable diseases (36%). In pts with disease control, the median PFS time was 3.8 months and the 3and 4.5-month PFS rates were 67% and 42%, respectively. With a median follow-up of 18.4 months (95% CI, 18.318.5), the median OS time was 13.1 months (95% CI, 8.7-17.5). The proportions of pts alive at 18 and 24 months were 33% and 25%, respectively. Pts who achieved disease control had a median OS of 15.4 months, while those who did not had 9.3 months. Median OS in pts pretreated with <3 and ≥3 regimens were 18.6 and 9.3 months (p=.03), respectively, whereas 1-year survival rates in biological-naive and priorbiological pts were 72% and 41% (p=.01), respectively. There was no toxicity exacerbation using the weekly schedule. In this cohort, two patients (15%) had PFS of 10.5 and 11.0 months, which resulted longer than PFS on prior therapy (3.8 and 6.3 months, respectively). Conclusion: Based on favourable toxicity profile and disease control in heavily pretreated CRC patients, NGR-hTNF will be further developed in combination with standard chemotherapy. Preclinical models have shown that tumour necrosis factor-alpha (TNF-α) has potent antitumour activity. However, its clinical use has been hampered by severe systemic toxicity, with MTD significantly lower than ED in humans1 NGR-hTNF is a novel vascular targeting agent (VTA) that has been rationally designed and prepared by coupling the N-terminus of human TNF-α with the C-terminus of the tumourhoming peptide Cys-Asn-Gly-Arg-Cys (CNGRC) (Figure 1) The cell surface receptor for the NGR-containing peptide is a CD13/aminopeptidase N isoform selectively expressed by endothelial cells of newly formed human tumour vessels2-4, including CRC (Figure 2) NGR-mTNF was found to have antitumour activity in preclinical model4 (Figure 3) even at doses in the picogram range (equivalent to a dose of 0.2 μg/m2 in humans) In a phase I study evaluating a dose-interval ranging from 0.2 to 60 μg/m2, the MTD of NGRhTNF was established at 45 μg/m2 when given as single agent once every 3 weeks5 A further trial exploring the low-dose range of NGR-hTNF from 0.2 to 1.6 μg/m2 selected the dose of 0.8 μg/m2 as the optimal biological dose, based mainly on dynamic imaging changes and preliminary antitumour activity6 Despite recent advances in metastatic CRC treatment, which include irinotecanor oxaliplatin-based first-line regimens, and increasing use of targeted agents, most patients develop resistance to these therapies Recently, two monoclonal antibodies have shown to be effective in unselected disease refractory to standard chemotherapy7,8 A significant increase of median PFS versus BSC was reported for patients treated with cetuximab7 (1.9 vs 1.8 months) and with panitumumab8 (2.0 vs 1.8 months). Similar results were registered in terms of disease control rate (39% vs 11%, in the first study and 37% vs 10%, in the second study). However, there is a need for new treatment options in this setting Results Table 3. Best overall response